Request rerouting for ISP operated CDNS

Size: px
Start display at page:

Download "Request rerouting for ISP operated CDNS"

Transcription

1 SCHOOL OF ELECTRICAL ENGINEERING AND TELECOMMUNICATION Request rerouting for ISP operated CDNS by Jeffrey Lai Thesis submitted as a requirement for the degree Bachelor of Engineering (Electrical Engineering) Submitted: November 3, 2014 Supervisor: Tim Moors Student ID: z Topic ID: TM125

2 Abstract Content Delivery Networks(CDN) currently utilise the Domain Name System to direct end users towards optimal CDN replica servers. However, current CDN redirection mechanisms do not adequately take into account DNS caching performed by end users. Specifically, end-users may cache DNS answers that later lead to unoptimal CDN nodes, possibly leading to a decreased browsing experience. This document proposes a CDN redirection mechanism for ISP operated CDNs that overcomes this problem. By utilising SDN techniques it is possible for ISPs to perform end-user redirection at a TCP flow-based level, without relying on the end-user having up-to-date DNS information. This method is shown to demonstrate improvements in several web-browsing performance parameters such as access-delay time and page-load time.

3 Thesis title: Request rerouting for ISP operated CDNs Topic number: TM125 Student Name: Jeffrey Lai Student ID: z A. Problem statement ISP operated CDNs and CDN-ISP collaborations are becoming increasingly common. These CDNs are in a unique position to be able to apply better request rerouting mechanisms, as well as utilise cloud and virtualised servers to rapidly create additional infrastructure as needed. There are currently no request rerouting mechanisms that can transparently reroute a user from one CDN node to another, if they have stale DNS information. Stale DNS information may arise when users cache DNS information, or if network topology changes occur that render their current information stale. B. Objective Perform an analysis on the effects of client side DNS caching on current DNS-based redirection mechanisms Design and implement a request rerouting mechanism that can reroute users in case of stale DNS information, as well as being able to utilise the creation of new CDN replica servers via cloud services. C. My solution DNS request rerouting mechanism already exists, and is proven to work Why change what works? Propose that we augment the current DNS request rerouting mechanism with SDN techniques Using OpenFlow flow rules, detect if users are going towards an IP address belonging to a CDN we know about. Check if these users are going to the ideal CDN replica server mapped for their location (That we know since we are their ISP) If they are not going to the optimal CDN node, then perform a DNAT on their outgoing packets, and a SNAT on packets coming back from the CDN If they are going to the right CDN node, then forward and route as usual D. Contributions (at most one per line, most important first) We have successfully created a request rerouting mechanism that augments the current DNS redirection mechanism and achieves the goals listed above. Performed a first of it's kind analysis on the effects of client side DNS caching on CDN performance. E. Suggestions for future work Analyze the scalability of the mechanism that has been proposed by this thesis Generate a request rerouting algorithm that works efficiently with the mechanism proposed by this thesis While I may have benefited from discussion with other people, I certify that this thesis is entirely my own work, except where appropriately documented acknowledgements are included. Signature: Jeffrey Lai Date: 30 / 10 / 14

4 Thesis Pointers List relevant page numbers in the column on the left. Be precise and selective: Don t list all pages of your thesis! 14 Problem Statement 14 Objective Theory (up to 5 most relevant ideas) 10 Current CDN redirection mechanisms have limitations 12 Software defined networking allows me to achieve my design goals 15 Analysis of possible ways to achieve my design goals 17 The design and theory behind my chosen solution Method of solution (up to 5 most relevant points) 19 Description of my solution 20 Description of the setup of my solution 21 Description of the testing framework for testing my solution Contributions (most important first) 35 Demonstrate that my controller works to the design goals 28 Create testing procedures to demonstrate the effects of DNS caching on CDN performance My work 21, 28 Testing procedures described, testing framework described 41 Appendix contains my code for my controller Description of procedure (e.g. for experiments) Results 31 Succinct presentation of results 35 Analysis 35 Significance of results Conclusion 38 Statement of whether the outcomes met the objectives 38 Suggestions for future research Literature: (up to 5 most important references) 3 Pushing cdn-isp collaboration to the limit 5 The akamai network: A platform for highperformance applications

5 Abbreviations API Application Programming Interface AS Autonomous System BGP Border Gateway Protocol CDN Content Distrubution Network CLI Command Line Interface DNS Domain Name System EDNS Exenstion mechanisms for DNS HTTP HyperText Transfer Protocol IP Internet Protocol ISP Internet Service Provider Mbps Megabits per second NAT Network Address Translation OS Operating System PoP Point of Presence QoE Quality of Experience QoS Quality of Service REST Representational State Transfer RTT Return Trip Time SDN Software Defined Networking TCP Transport Control Protocol TTL Time To Live URL Uniform Resource Locator 3

6 Contents 1 Introduction 8 2 Background Content Delivery Networks ISP operated CDNs Request rerouting Current request rerouting mechanisms Limitations of current request rerouting mechanisms Software Defined Networking Literature review Request Rerouting Mechanisms for ISP Operated CDNs Problem statement & design goals Analysis of possible mechanisms Anycasting CDN nodes Using DNS augmented with NAT Experimental set up of a DNS request rerouting mechanism augmented with NAT Testing framework Testing procedures Results & Performance Request rerouting after link failure Request rerouting during link congestion Request rerouting with the introduction of a new CDN replica server Evaluation Results Discussion Conclusion Future Work Bibliography 42 4

7 Appendix An example of a DNS request for facebook.com, a CDN hosted website HTML for wikipedia.com The Ryu controller for my thesis The mininet topology for my thesis

8 List of Figures 2.1 A visit to yields HTTP requests to various CDN hosts, as seen in the left-hand column Our network topology diagram A Trace demonstrating the typical NAT procedure that happens to a TCP flow A wireshark trace of a DNS request for facebook.com

9 List of Tables 2.1 Browser usage by market share Design requirements fulfilled by utilising an IP Anycast method Table miss flow that catches everything. Used as part of learning switch functionality Table miss flow that catches traffic from h1( ) to the CDN replica servers An example forwarding flow inserted into the switch in the case that there is no need to further redirect the end user. Note that IP destination and TCP source port values are context dependant, the values here are just examples Example flows inserted into the switch in case there is a need to further redirect the end user. Note that as with the previous table, the IP address and TCP port values are arbitrary in this case, but remain relevant between the DNAT and SNAT rules Time taken to fully traverse and load the thesis.com testpages, in seconds Time taken to fully traverse and load the thesis.com testpages, in seconds. Also shown is time delay until the client connects to the optimal CDN server over the congested one Time taken to fully traverse and load the thesis.com testpages, in seconds. 36 7

10 Chapter 1 Introduction Content Delivery Networks are finding an increase in usage in today s internet landscape. In response to a growing demand for high definition content, Internet Service Providers are also starting to provide and host their own CDNs [1, 2]. Large CDN providers such as Akamai have also started collaborating with ISPs to provide better services to their end users [3]. These ISP operated CDNs are in a unique position wherein they control the edge routers that sit at the last hop to their users. Because of this, ISP operated CDNs 1 have a larger amount of control over how their end users are redirected towards CDN nodes. Currently, hostnames for CDN hosted websites are resolved using DNS services that provide end users with an IP address that maps to the optimal CDN node for that user at the time of request [4]. However, end user caching of DNS information can lead to an end user accessing a nonoptimal CDN node. To illustrate, we show in this thesis how a website s Quality of Experience is affected by end user caching of DNS information in modern web browsers and operating systems. The results of this testing indicate that indeed, end user DNS caching has a negative effect on website QoE. As such, DNS caching also inhibits the rapid startup and application of cloud services to adapt to changes in network topology and congestion, possibly inhibiting future ISP CDN collaboration [3]. By utilising newly emerging Software Defined Networking techniques, we demonstrate that it is possible for ISP operated CDNs to overcome the issues caused by end user caching of DNS information. By using a topologically aware network controller, NAT rules are applied to end user TCP flows, providing transparent redirection in cases where users are utilising stale DNS information. This method is applicable to not only wholly owned ISP CDNs, but also for ISP CDN collaborations. In summary, the contributions of this thesis to the area of CDN redirection are: We demonstrate the problems associated with client side caching of DNS responses to requests for CDN information. We design and implement a novel redirection mechanism for ISP operated CDNs, involving the use of newly emerging Software Defined Networking techniques to augment traditional DNS mechanisms with NAT operations. 1 For the purpose of this document, the term ISP operated CDN refers to both CDNs wholly owned and run by an ISP, as well as ISP CDN collaborations. 8

11 Chapter 2 provides the background and literature review for this document. Chapter 3 explores possible request rerouting mechanisms available to ISP operated CDNs, and describes the method and testing procedure used for the request rerouting mechanism proposed by this thesis. Chapter 4 shows the results from testing and executing the proposed CDN redirection method. Chapter 5 evaluates and discusses these results. Chapter 6 provides a summary of the findings of this thesis, and suggests areas of future research and development. 9

12 Chapter 2 Background 2.1 Content Delivery Networks CDNs have become the primary providers of content in today s internet environment [5, 6]. With rapid changes and increases in end user content consumption, CDNs have been increasing their network footprints to keep up with traffic requirements. Current CDNs aim to serve content from content providers to end users in an efficient manner. While the principles of operation between different CDNs may differ, their overall structure is the same: replicas of content are pushed from the content provider s origin server to replica servers across the CDN s network. When an end user makes a request for that CDN hosted content, it is served by a replica server belonging to the CDN. This replica server is determined by a variety of factors such as current network load, content availability, geographical proximity, and latency. This provides many benefits for both the end user and the content provider, such as load-balancing and timeliness of delivery [7]. Primarily, an end user accesses CDN hosted content by visiting a website. With current CDN models, websites are either completely hosted by the CDN, or contain links to CDN hosted content 1. CDNs are now used to host a majority of content on the web. In 2012, a survey of the top 1000 most popular sites was performed by John et al. [4], which revealed that all the sites in the top 10 and over 70% of those in the top 1000 relied on CDNs. With the prevalence of cross-service APIs and resources, and links for sharing content onto social media sites such as twitter and facebook, a single site can create HTTP requests for multiple CDNs. This can can be seen in figure 2.1 below, where requests for not only abc s CDN are made, but also CDNs for advertising, social media, and analytics networks as well. 2 1 For example, text on the website may be hosted on the origin server, but URLs for images and videos may lead to CDN replica servers. 2 These do not necessarily go to individual CDN providers. static.ak.facebook.com and cdn.media.abc.go.com are both served by Akamai, for example. 10

13 Figure 2.1: A visit to yields HTTP requests to various CDN hosts, as seen in the left-hand column ISP operated CDNs As user demands for internet hosted content continue to grow [8, 6], ISPs have started to roll out their own CDNs to keep up [2, 1]. Previously, ISP operated caches were operated for the same effect [9], but not to the extent and sophistication of a CDN. Aside from customer Quality of Experience, there is a networking motivation for ISPs to operate their own CDNs, such as having lower inter-networking traffic, thus reducing the cost and requirements of building and maintaining inter-network infrastructure [10]. ISPs have also started collaborating with larger commercial CDN providers such as Akamai [3], and Netflix [11], to achieve the same benefits described above. ISP operated CDNs have various qualities that make them distinct from other CDN providers. Because ISPs operate as the last mile network for their customers, they can control the specifics of how their clients are routed towards CDN replica servers with a higher level of control than other CDN providers. Also, being the network operator, they have access to network statistics that are otherwise unavailable to other CDN providers, such as link congestion, peak traffic, and the exact network location of their customers. Because of these qualities, ISP operated CDNs can utilise request routing mechanisms 11

14 that are otherwise unavailable to traditional CDNs. These ISP operated CDNs can also better utilise technologies such as cloud services to rapidly and dynamically create replica servers as demand requires [3]. 2.2 Request rerouting In order for CDNs to provide clients with the content they request, clients must first be directed to the server most appropriate for them. This is referred to as Request Rerouting. A request rerouting system is composed of two parts - A request rerouting algorithm that is used to decide which server address to provide for user requests, and a request rerouting mechanism that actually provides users with said addresses [12]. This thesis focuses on the request rerouting mechanisms for ISP operated CDNs Current request rerouting mechanisms Augmented DNS While multiple approaches to CDN request rerouting currently exist, companies predominantly use augmented DNS servers, or HTTP-based redirects [12, 13]. CDNs that use DNS based request rerouting mechanisms utilise augmented DNS servers that act as the authoritative domain for their customers 3. With this method of request rerouting, when an end user s browser visits a website hosted by the CDN, it must first make a DNS request for the IP address corresponding to that hostname. For example, when an end user wishes to visit their browser first makes a DNS request in order to create an IP address/hostname mapping 4. When the authoritative DNS name server for that website receives the request 5, it uses parameters such as latency, server load, and the geographical location of the end user to determine which replica node can provide the best QoE to the end user. By altering the responses given by their DNS servers, CDNs utilising this request rerouting mechanism can choose which content servers provide content to end users URL redirection DNS is also used in a request redirection mechanism known as URL redirection. In this mechanism, when a browser initially resolves a URL, they make an initial GET request towards the IP address that maps to that URL. The server at that IP address then replies in a manner that causes the client to be directed to another URL either by a HTTP status code, Javascript redirect, or other methods[14]. For example, when a user visits on their browser, they are provided with a HTTP status 301 Moved Permanently status code, along with a URL pointing to 3 Akamai is an example of a CDN provider that does this. [5] 4 A wireshark trace of a DNS request chain for htpp:// has been provided in the appendix. 1 5 This assumes that the client s original DNS resolver did note return a cached result. More detail is provided in Section

15 and the browser automatically redirects to the new URL. Unfortunately, this method incurs extra access delay time for each element that is accessed, as the initial GET request is essentially wasted because it only serves to provide a redirect towards another URL. Additionally, another DNS request has to be performed to resolve the IP address of this new URL 6. This is okay for large blocks of content such as videos as it only incurs a once off delay, but undesirable for smaller content such as images IP Anycast Another request rerouting mechanism involves allocating an anycast IP address [15] to multiple CDN replica servers. In this mechanism, all DNS servers reply with the same IP address for CDN content servers. When a client makes a content request to using that IP address however, they are then routed to the closest CDN replica server due to the nature of IP Anycasting Limitations of current request rerouting mechanisms While current mechanisms of request rerouting work, they invariably suffer from inherent limitations due to both the nature of their implementation, and a lack of network metrics available to traditional CDN operators [3] DNS caching Specifically, a DNS based request rerouting mechanism relies on end users actually making DNS requests, in order to provide them with an IP address mapping to an optimal CDN replica server. While this is fine in practice, it does not take into consideration caching of DNS information by end users. For example, if a user visits a CDN hosted website and caches the DNS information provided by the CDN s DNS server, then future requests for content hosted on that website will not generate a DNS request until the cache entry times out. This can cause issues if a more optimal CDN replica server becomes available during the meantime. Currently, CDN operators assign very low TTLs to their DNS entries in order to work around this [5], typically under a minute 7. However, nearly all internet users are currently in situations where these TTLs are not honoured: In September 2014, the w3schools organisation reports these statistics [16] on browser usage: 6 The HTML for is provided in the appendix.2 7 For example, performing a dns request for akamai.com reveals a TTL of about 20 seconds on their responses 13

16 Out of these browsers: Browser Market share Chrome 59.6% Firefox 24.0% Internet Explorer 9.9% Safari 3.6% Opera 1.6% Table 2.1: Browser usage by market share Chrome caches DNS entries for 60 seconds or the DNS TTL, whichever is lowest [17] Firefox caches DNS entries for 60 seconds [18] Internet Explorer caches DNS entries for 30 minutes [19] Opera caches DNS entries for 10 minutes [20] From the referenced technical documentation, it is evident that the vast majority of DNS requests for CDN hosted content have their responses cached. If there are any other possible DNS caches that are being utilised, this amplifies the effect even further. Chrome, Firefox, and Internet Explorer also perform link prefetching [21, 22, 23], where DNS requests are made before a user actually follows those links. A browser achieves this by looking at the links on a page and performing the DNS requests for their hostnames. These browsers will also perform DNS requests during startup, to reduce the overall load time of a user s most commonly accessed websites. The above two effects of DNS caching and link prefetching compound together to create the following possible scenario: It is possible for the IP address of a hostname to change, while an end user is still using the old IP address. For example, if a user launches their browser, generating a few DNS requests via link prefetching, and only visits the websites that were prefetched a minute later, it is possible that the IP address they are using is stale. This can lead to QoE degradation for the user: Content may be out of date, or the CDN replica server may be congested, for example. Because of this, any serious attempt at improving the CDN redirection mechanism has to take into account user end DNS caching and link prefetching. Failure to do so implies that any changes to the CDN can take up to a minute or more for its effects to be felt for all users of the CDN Problems with anycast The anycast method for directing users to CDN nodes, while largely immune to the problems with DNS caching as described above 8, still suffers from issues that make it inappropriate for ISP operated CDNs. Because IP anycasting routes users to the closest network that advertises a BGP route for that IP address, it fails to take into consideration 8 Since DNS requests for anycasted CDN nodes all point to the same anycast IP address, these IP addresses rarely, if ever, change. 14

17 other parameters that are important for a good QoE, such as network congestion. This can lead to additional network bottlenecks [3], and cause extra congestion on already congested links. In order for current anycasting mechanisms to redirect clients from a congested server to another, less congested server, they must either be redirected from their originally requested server (Which adds to server load and latency), or changes to BGP routing tables must occur, which can take a relatively long time to propagate across a network [15]. Additionally, the configuration of an anycast mechanism can be different for each network. BGP routes have to be configured, and available IP spaces considered. Indeed, even industry-leading CDN operaters such as MaxCDN [24] and Cloudflare [25] claim It is not easy to setup a true Anycasted network and that IP anycasting is complex 2.3 Software Defined Networking Software defined networking is a new networking paradigm that decouples the control plane in a network from the data plane. Instead of routing decisions being made by a layer 2/3 device, these decisions are instead made by a controller that has a view of the entire network. This overarching controller then installs flow rules into switches on the network, which allows it to perform networking functions similar to current networks. This allows network operaters to make rapid, adaptable changes to a network as needed. Currently, the most popular SDN protocol used for this purpose is OpenFlow [26], which defines the communication between an SDN controller and the switches that it governs. By using OpenFlow, network researchers can perform networking experiments and achieve network functionality in an easily reproducible and flexible manner. Because OpenFlow provides software abstractions and APIs for hardware functions, it becomes easier to programmatically apply network rules and functionality from a centralised location. In this thesis, SDN and OpenFlow is used to dynamically create the redirection rules that direct end users to more optimal CDN nodes. A controller using the open source Ryu framework [27] communicates to an OpenFlow capable switch to push flow rules as needed. 2.4 Literature review Much work has been done on CDNs [5, 7] in the last decade, due to the large effects they have on increasing the effectiveness of content delivery. However, despite the fact that DNS caching has been around since CDNs were first established, published research on the effects of DNS caching by end users has not yet been explored in depth. A recently established working group in the IETF are developing a set of standards that define interconnections and collaborations between CDNs called the CDNI [28]. Work has been done within this set of standards to provide CDNI request routing using SDN methods [29]. However, it focuses on the request routing between upstream and downstream CDN providers in a CDNI, and not between an end user and CDN replica nodes. As discussed in section , a current request redirection mechanism for CDNs is IP anycasting. Current research by Yulei Wu et al. [30] demonstrates an SDN implemen- 15

18 tation of IP anycast that overcomes IP anycast s main issue of being unaware of network conditions such as congestion, a so called load-aware anycast solution. While this implementation would work fine for a CDN that is wholly owned and operated by an ISP, it remains inflexible and unscalable for ISP CDN collaborations. Further analysis on this matter is provided in Section OpenCache [31], an SDN caching platform, has recently been developed and is currently under research. However, it is more focused on providing a caching solution for high quality streaming video. They claim that their OpenCache system can be used to enhance CDN ISP collaborations though, and indeed, work from this thesis could be combined with OpenCache to provide tools for a unified CDN collaboration/isp CDN framework. Also, their implementation involves flow granularity at the IP level, whereas our request rerouting mechanism is based off TCP sessions. Work done by Akamai[3] on CDN ISP collaboration has shown that a DNS based system can adequately perform user server assignment in ISP CDNs. It also discusses the viability of rapidly spinning up virtual cloud servers for the purpose of acting as CDN replica servers in less than a minute. This confirms that having a request redirection mechanism that can actively redirect users to more optimal CDN replica nodes in under a minute is a desirable outcome to achieve. An example of an existing ISP operated CDN is one operated by Comcast [32], where they use a combination of DNS and URL redirection techniques to direct their customers to suitable caches for their requests. However, their DNS mechanism does not avoid the limitations associated with user side caching. Also, their network is focused on utilising static caches in pre-established locations, with no mention on the possibility of utilising a virtualised replica node. 16

19 Chapter 3 Request Rerouting Mechanisms for ISP Operated CDNs 3.1 Problem statement & design goals In order for a request rerouting mechanism to be successful, it should fulfill the following criteria: Be transparent to the end user and require no extra configuration client-side Successfully provide clients with addresses to content servers Provide clients with access to optimal servers for their content requests Be failure resilient These are the base requirements that all current request rerouting mechanisms aim to achieve [12, 3]. Also, as detailed in Section 2.2.2, current CDN mechanisms: Are unable to provide rapid server changes to end users due to the effects of DNS caching (In the case of augmented DNS) Are unable to fully utilise network information available to a network operator (In the case of IP anycasting) Require complicated configuration to set up (In the case of IP anycasting) As such, any request rerouting mechanism created as the result of work performed under this thesis should also fulfill the following requirements: In the case of DNS caching and/or prefetching and network changes that may happen during that caching time-frame, be able to continually and transparently reroute users to the optimal CDN replica node for their request. Be able to fully utilise information available to the network operator 17

20 Minimise the configuration necessary to implement our proposed mechanism, so as to maximise ease of ISP-CDN collaboration and ISP-CDN establishment. Because the scope of this thesis is to provide a request rerouting mechanism for ISPs, it is also desirable to define a set of design goals that enable a simpler, more attractive implementation of our proposed mechanisms for ISPs: Be incrementally implementable, to lower implementation costs and reduce network downtime Be able to utilise and act on the ability to rapidly create virtual CDN replica servers using cloud services 1 Be scalable for network and service expansion. Optionally, be able to provide access to the CDN for end users that are not customers of the ISP. These requirements will be used as a guideline throughout the rest of this document in design, execution, and results evaluation. 3.2 Analysis of possible mechanisms In this section, an analysis of possible methods of performing request rerouting for ISP CDNs is performed. While a NAT-based augmentation of DNS services is the method proposed by this thesis, a detailed justification of this proposition is provided Anycasting CDN nodes One of the initial considerations when beginning this thesis project was to utilise IP anycasting to perform CDN rerouting. However, further analysis of this mechanism revealed that it would be unsuitable for our required goals. What follows is the analysis that led to this decision: Anycasting allows for a one-to-many association of a hostname to CDN replica server endpoints. That is, when a browser makes a DNS request for CDN hosted content, they will always receive an anycasted IP address that leads to the closest replica server that responds to that address. This makes the issues caused by DNS caching non-existent, as the IP address in an anycasting setup should ideally never change. However, this is also a drawback; because IP anycasting only directs requests towards the closest server replica, it does not consider server load. Benefically however, this also means that clients using remote DNS services 2 will still get directed to a CDN replica server close to them without requiring the use of EDNS. 1 As per recommendations in Akamai s ISP collaboration paper[3] 2 Services such as OpenDNS, as opposed to the DNS servers provided by an ISP 18

21 Additionally, the nature of IP anycasting creates a potential to pollute the global routing system, as it relies on BGP to function. This can lead to major network incidents that affect uptime: An example being in October of 2014 when a Latin American ISP accidentally directed traffic originally destined for Cloudflare to concentrate on a single data center, quickly overwhelming it [33]. This incident was caused by a false BGP announcement, where the Latin American ISP announced via BGP that their network, instead of ours, handled traffic for CloudFlare. It is important to note that failures of this type leads to a complete loss of access to the CDN datacenter throughout the duration of the failure. In this incident, access to Cloudflare s services was lost for 49 minutes in not just Latin America, but also other parts of the world: Traffic to CloudFlares customers sites dropped by 50% in North America and 12% in Europe. This is a catastrophic failure for a service that aims to increase the availability of content hosted by it. This reliance on BGP means that an IP anycasted network will also require custom configuration and set up per network. As stated in industry blogs for leading CDNs that utilise IP anycasting themselves, this configuration can be complex and hard to setup [24, 25], which may be offputting for ISPs looking to lower costs. With an increasing trend in utilising cloud services [3], the use of cloud services to rapidly establish CDN replica servers is becoming a distinct possibility. This can cause issues with IP anycasting however; because IP anycasting directs packets to the nearest server that responds to that anycasted IP address, this will cause a newly created server to suck in requests from nearby clients. For example, if a client is streaming video from server A that is 10 hops away, and then server B is populated at a location 7 hops away, then further requests from that client will be forwarded to server B, causing their video streaming session to close suddenly, decreasing QoE for the client. Past research by Hitesh Ballani et al. [34] indicates that this behaviour is to be expected in current anycast systems. While current CDN operators that implement IP anycast may have techniques to avoid this form of route flapping, they are currently unavailable to the author, and so will not be considered for this facet of IP anycasting. In summary, IP Anycasting in it s current state has the following advantages and disadvantages 3 : Advantages: Disadvantages: Not affected by client-side DNS caching Not affected by the use of remote DNS servers Catastrophic failure modes 4 Unable to utilise cloud services due to route flapping Unable to utilise network information, such as congestion Requires a lot of configuration 4 Possibly affected by BGP pollution/route leakage 3 Ignoring the inherent advantages and disadvantages that come with using a CDN, such as increased availability of content 4 Compared to a DNS based mechanism, below 19

22 In summary, using an IP Anycasting mechanism fulfills the following design goals 5 : Able to work around DNS caching/prefetching! Able to fully utilise network information![30] Easy to configure # Incrementally implementable # Able to utilise and act on the ability to rapidly create virtual CDN replica servers # Scalable for network and service expansion! Able to provide access to the CDN for end users that are not customers of the ISP! Table 3.1: Design requirements fulfilled by utilising an IP Anycast method While an SDN implementation of IP anycast could work around a few of the disadvantages such as the inability to utilise network information [30], the other inherent disadvantages of IP anycasting heavily outweigh the advantages that IP anycasting provides. Namely, the inability to utilise cloud services to rapidly create extra network resources as necessary 6, and the heavy configuration requirements mean that IP anycasting in it s current state is not suitable for the rapidly expanding content delivery requirements [8] of ISP networks in the future, in the context of ISP-CDN collaborations Using DNS augmented with NAT The request rerouting mechanism proposed by this thesis utilises current DNS-based CDN rerouting mechanisms, but augments it with NAT to overcome the issues of client-side DNS caching. This subsection contains the analysis that led to the decision to use this method, as well as a detailed explanation of said method. An analysis on the worthiness of creating such a request rerouting mechanism is also explored. As mentioned in Chapter 2, a currently popular method for performing CDN rerouting is with augmented DNS servers. Current implementations of DNS-based rerouting mechanisms already fulfill most of the design goals mentioned in the start of this chapter; indeed, some current ISP CDN implementations utilise augmented DNS [32] to perform request rerouting, indicating that CDN rerouting with DNS is at least good enough for production ISP CDN networks. Knowing that DNS-based rerouting is a already a viable option for ISP CDN request rerouting, it is now important to determine if an improved request rerouting mechanism is worth both development and implementation in ISP operated CDNs. The main limitation with current DNS request rerouting mechanisms identified in this thesis is the fact that clients perform DNS caching that does not honour TTLs set by the CDN s DNS resolver. This limitation implies that for an ISP wishing to utilise rapid establishment of CDN replica servers, their users potentially will only see the improvements of such an action 5 Ignoring the base requirements needed for a successful CDN rerouting mechanism as outlined in the beginning of this chapter 6 Which Akamai encourages [3] for future CDN collaborations 20

23 minutes after it was put into effect. Similarly, this also means that it may take over a minute to simply reroute users to other CDN replica nodes that already exist in the network. This can potentially lead to windows where end users are not being provided with the best achievable QoS that the network can provide, despite those resources existing. Given that one of the aims for current ISP CDNs is to be able to deliver video content such as Live TV and Video On Demand [2, 32], lowered QoS attributes such as access delay time and bitrate can harshly lower customer satisfaction and engagement in ISP networks. A study done by Florin Dobrian et al. [35] indicates that a 1% increase in buffering ratio 7 can reduce user engagement by more than 3 minutes for a 90-minute live video event. As user expectations for network performance increase over the years[8], this will become an even more important factor in customer satisfaction. Scenarios where an ISP may want to rapidly establish extra CDN replica servers, or otherwise reroute users away from the node they are currently connected to include: Flash floods, where a sudden influx of users (say, to watch a live sports event) can potentially overload a CDN cache location Sudden link/datacenter failure Sudden availability of more current content Wanting to redistribute link congestion across the network (maybe to ensure SLAs for other ISP customers) As a real life example of the importance of being able to handle the first point above, an outage on the HBO GO network during a season premiere of Game of Thrones in April, 2014 affected nearly a million people [36]. The amount of public negativity that occcurred after this event demonstrates how lack of accessibility during key moments can have a large impact on a network s reputation 8. Thus, it would be ideal to be able to minimise the possible negative effects of flashfloods on CDNs. To that end, it is clear that implementing a CDN rerouting mechanism that can accommodate a rapid introduction of new CDN replica nodes has value in an ISP CDN. Research done by Akamai [3] indicates that it is possible to add, remove, and migrate live servers on demand in less than a minute, indicating that it would be possible for a ISP CDN network to overcome incidents like the one above by allocating extra CDN replica servers as demand increases. Recent work by Justine Sherry et al. [37] indicates that it is indeed practical and feasible for enterprise networks to outsource middlebox processing to the cloud. Also, they indicate that To allow redirection using multiple cloud PoPs, we can rely on DNSbased redirection similar to its use in CDNs So, knowing that: a) It is important to keep QoS high for the types of content that an ISP CDN is likely to provide; b) It is possible to keep QoS high by rapidly outsourcing cloud services to spin up Virtual Machines to run CDN replica servers; and c) DNS caching means that any changes to network topologies can potentially take over a minute to be 7 The amount of time a video is playing as opposed to sitting idle and buffering 8 This is becoming increasingly true with the advent of social media platforms such as twitter where disenfranchised patrons can vent into the public space 21

24 fully noticed by an end user, it is therefore both desirable and effective to have a request rerouting mechanism that can overcome the effects of DNS caching by end users. With current augmented DNS mechanisms for request rerouting being so widespread [12], and it s effectiveness in successfully rerouting users to CDN replica servers, a decision was made to develop a mechanism whereby it would be possible to additionally augment the current DNS rerouting methods in such a way that the issues associated with DNS caching could be avoided. As such, we propose a CDN redirection mechanism whereby request rerouting happens via the regular augmented DNS method. However, if an end user is determined by the network to be requesting content from an unoptimal node, that request is then transparently redirected to the optimal node for that end user. This redirection occurs at one of the ISP s edge routers that serves that user. This is achieved by using Software Defined Networking techniques to install the relevant flow rules that perform the redirection on these packets at the edge router. In this context, there is a network controller device with access to network statistics from the ISP CDN network. Using this information, decisions are then made on whether or not to push flow rules to certain edge switches to determine if they should perform redirection on packets that pass through them. Transparent user redirection is achieved by performing destination-nat on their outgoing packets, redirecting them towards the appropriate CDN replica server. For packets replying from the CDN replica server, they have a Source NAT applied to them at the edge router before they are sent to the end user. The controller logic can be explained with the following rules: The controller knows the location of all customers in the ISP network. Similarly, it also knows the location of all current CDN replica nodes. The controller therefore knows which CDN replica nodes are optimal for each set of customers that are served by an edge router in the ISP network. For example, CDN replica server A should serve the customers in the suburb of Ashfield, who are all connected to the rest of the ISP network via a single edge router at their local exchange. If and when network conditions change, the mappings set out above may also change. For example, CDN replica server A may be currently under high congestion, so the suburb of Ashfield should instead be directed to CDN replica server B. If and when these mappings of servers to customers changes, then the controller shall install routing rules into the edge routers that serve these customers, that transparently redirect them towards the more optimal server. That way, if they are using stale CDN information, they still get directed to the optimal CDN replica server for them. If they are using the new CDN information, routing happens as usual to the expected destination. Following this controller logic, the CDN request rerouting process for an end user in this scenario is outlined on the next page: 22

25 1. A client types a URL into their browser, wishing to view content that is hosted on the ISP CDN. 2. The client s browser performs a DNS request to map an IP address to the URL entered into the browser (a) Either this request is fulfilled by the browser/os DNS cache, and a result is returned instantly, or (b) The requested URL is not in the cache, and is resolved by an external DNS server. 3. The client s browser creates a new TCP session to communicate with the CDN replica server. This session can be identified by a source and destination IP, and source and destination TCP ports (In the case of HTTP, destination ports are always 80 or 443) 4. The initial packet for this TCP session reaches an edge router belonging to the ISP. (a) If the IP destination for this TCP session is the same as the optimal CDN server that has been mapped for this customer, then no redirection needs to occur. A new flow that routes this packet towards the CDN server is created, and the packet is routed as per normal. (b) If the IP destination for this TCP session does not match the IP of the optimal CDN server that has been mapped for this customer, then redirection needs to occur. A new flow is created that a) Performs a Destination based NAT, changing the destination IP address of packets that belong to this TCP session from the old CDN server address, to the new optimal CDN address; and b) Also routes the packet towards the optimal CDN node, as in a regular router. A reverse flow is also created, that performs source based NAT on packets returning from the CDN replica server. Because new flows are created only when new TCP sessions pass through ISP edge router, scenarios where end user TCP flows are suddenly and unexpectedly terminated are thus avoided. 23

26 3.3 Experimental set up of a DNS request rerouting mechanism augmented with NAT In this section, we describe the framework and programs that were used to create, implement, and test the request rerouting mechanism outlined above Testing framework The network topology We use Mininet [38], a network virtualisation tool, to create a virtual network topology in which we perform our testing. Mininet provides an API that allows programmers to quickly and easily create a network topology, populate it with hosts, switches and controllers, and run network tests. A copy of the source code used to create the topology is included in the Appendix.4. In our tests with mininet, we also use Open vswitch 2.02 for our virtualised switches. In our networking tests, our topology contains the following devices: h1: A host used for the purposes of testing our improved CDN rerouting mechanism. They act as our host that exists within the ISP s network. It has a 20Mbps connection to s1 to simulate an ADSL2+ high speed subscriber line. h2, h3: Webservers acting as our CDN replica servers. They both serve identical websites and content. When uncongested, they have a 100Mbps connection to s1. When we are simulating them as being congested, we lower that connection speed down to 1Mbps. h4: A host used for the purposes of contrasting standard CDN rerouting mechanisms compared to our improved mechanism. They act as our host that exists outside the ISP s network, and thus does not receive the effects of our proposed request rerouting mechanism. It has a 20Mbps connection to s1 to simulate an ADSL2+ high speed subscriber line. h5: A DNS server s1: An OpenFlow enabled switch c0: Our Ryu controller. All links have been configured to add a 10ms delay to packets passing through them to simulate network locations that are all close to an ISP edge router. This network topology was run under the Ubuntu operating system. It is important to note that the Ubuntu operating system does not operate an inbuilt DNS cache. Figure 3.1 shows how our topology was linked together. 24

27 Figure 3.1: Our network topology diagram The SDN controller Ryu [27] was selected as the controller framework for this thesis. It was mainly chosen because the author was proficient with python, the language that the Ryu framework uses. This framework provides a rapid prototyping and testing platform for testing SDN OpenFlow controller applications, and an in-built REST API. We use the OpenFlow1.3 spec to communicate between our controller and network switch. While OpenFlow1.4 spec is the most current version as of writing, Ryu s last stable release during the course of this thesis supported OpenFlow1.3. A REST API is implemented via shell scripting to notify network changes to the controller. The source code for our controller, ThesisController.py, can be found in the Appendix.3. On controller startup, it installs two table-miss flow rules to the switch: One that matches on all traffic, with priority 0(As seen in Table 3.2), and one that matches on TCP HTTP traffic coming from h1( ) and headed towards the CDN subnet( /24) 9, with priority 10(As seen in Table 3.3). If a packet matches the rule with priority 0, then that implies there are no other rules on the switch that match that packet. This causes the switch to forward the packet to the controller for processing. 9 This subnet could be any size/value, we have just set it to /24 for convenience 25

28 When this packet arrives at the controller, it s source MAC address is mapped to it s ingress port, and if it s destination MAC address is known to the controller, then a flow rule is created that maps this particular source MAC/Dest MAC pairing to prevent further packet-ins. Otherwise, the packet is flooded out from the switch. In this manner, our controller causes the switch to act exactly as a learning switch. Priority Ingress Proto- IP Source IP Desti- Source Destination Actions Port col Address nation Port Port 0 * * * * * * Send packet to controller Table 3.2: Table miss flow that catches everything. functionality Used as part of learning switch Priority Ingress Proto- IP Source IP Desti- Source Destination Actions Port col Address nation Port Port 10 * TCP /24 * 80 Send packet to controller Table 3.3: Table miss flow that catches traffic from h1( ) to the CDN replica servers. The second flow table miss entry however, matches on HTTP traffic with source IP of h1( ), and destination IP /24. Traffic that matches this rule at the switch is also sent to the controller. Now, depending on the controller and network state, one of two things happens: If there have been no changes to the network topology, then h1 s traffic that is directed towards the CDN should have a destination IP address that maps to the ideal CDN node for that customer. In this scenario, a regular flow(as seen in Table 3.4) rule is created that matches on the HTTP TCP session of the packet that was sent to the controller. This means any further packets from h1 to the CDN that belong to this TCP session will not hit the table-miss flow, and will get forwarded normally. Ingress Port Priority Protocol IP Source Address Source Port IP Destination Destination Port Actions 20 * TCP Send out on switch port corresponding to destination IP address Table 3.4: An example forwarding flow inserted into the switch in the case that there is no need to further redirect the end user. Note that IP destination and TCP source port values are context dependant, the values here are just examples. If there have been topology changes however, and h1 is sending packets to a nonoptimal CDN replica server, then the controller detects that the destination IP of this packet does not match the IP address of the optimal CDN replica server for h1. In this scenario, a flow 26

29 that performs a DNAT action on the packets belonging to this TCP session is created. Correspondingly, a flow that performs an SNAT action on packets belonging to this flow is also created. These flows can be seen in Table 3.5. In order to notify the controller of these topology changes, a REST API is used to update the controller with the current network status. Ingress Port Priority Protocol IP Source Address Source Port IP Destination Destination Port Actions 20 * TCP DNAT destination IP from to , Send out on switch port corresponding to destination IP address 20 * TCP SNAT source IP from to , Send out on switch port corresponding to destination IP address Table 3.5: Example flows inserted into the switch in case there is a need to further redirect the end user. Note that as with the previous table, the IP address and TCP port values are arbitrary in this case, but remain relevant between the DNAT and SNAT rules. 27

30 A trace of typical HTTP TCP sessions under this request rerouting framework is provided below in Figure 3.2. Figure 3.2: A Trace demonstrating the typical NAT procedure that happens to a TCP flow. The packets are as follows: 1. DNS request for thesis.com 2. DNS response for thesis.com, IP of TCP-SYN from to as seen from h1 s point of view 4. TCP-SYN from to , after it has been DNAT d at the switch s1 (Packet now TCP-SYN from to ) 5. TCP-SYN, ACK from to before SNAT 6. TCP-SYN, ACK from to after SNAT (Packet now TCP-SYN, ACK from to ) 7. the packets afterwards follow the same pattern of DNAT, and SNAT. 28

31 These flow rules that are installed by the switch therefore redirect users on a TCP session based level. This way, end users are not suddenly disconnected by a TCP session being redirected mid-flow. More analysis on this facet of our request rerouting mechanism is provided in Chapter 5. The code from our Ryu controller is provided in the Appendix.3. It uses some code snippets from the open source Ryu examples OpenFlow enabled switch The switch is an instance of Open vswitch run in the Mininet topology. We can inspect the flow tables of this switch at any time by issuing commands at the CLI. The switch is made to run OpenFlow1.3 spec. It operates as a learning switch with some extra flow rules that allow it to perform NAT and DNAT operations on TCP flows that pass through it DNS server Because this thesis is primarily concerned with the request rerouting mechanism, and not the request rerouting algorithm [12], we keep our DNS server simple. It is implemented on h5 via the BIND DNS Nameserver [39]. When we want to simulate a network topology change(the addition or changing of the optimal CDN replica server that maps to the hostname), we update the A record provided by h5 via a shell script. In our testing framework, this DNS server acts as the authoritative nameserver for thesis.com. This server has an IP address of , and sets the TTL on it s responses to be 30 seconds CDN replica servers As above, because we are primarily concerned with the request rerouting mechanism, we keep our web servers simple. Our webservers at h2 and h3 both point to the same directory to serve their webpages; the content they serve is exactly identical. This is functionally similar to a CDN provider that has replicas of content across multiple servers. We run our webservers using a simple python script that serves files from the current working directory on port 80. h2 and h3 have IP addresses of and , respectively. They both serve as the webserver for thesis.com. thesis.com consists of a landing page with links to other articles, and a link to thesis.com/firstpage.html, which is 36megabits in size. This page links to thesis.com/secondpage.html, which is 31megabits in size. Finally, this page links to thesis.com/thirdpage.html, which is 40megabits in size. The content on each page consists of 10 high resolution pictures. There are also a series of javascript scripts on each page, for the purposes of automating the actions of the end users h1 and h4 that visit it. The landing page itself has a script that waits for 10 seconds, and then directs the browser to thesis.com/firstpage.html. On thesis.com/firstpage.html, there is a script that loads thesis.com/secondpage.html, after thesis.com/firstpage.html has been fully loaded. Similar behaviour occurs on secondpage.html, and thirdpage.html. thirdpage.html leads back to the thesis.com landing page, allowing the cycle to continue again. 29

32 The end users h1 and h4 act as the end users in our network. They are just simple mininet hosts running web browsers. In our tests, two browsers are used: Mozilla Firefox, and Google Chrome. This should provide a fair representation of a large proportion of the current browser market share - w3schools estimates 83.6% of internet users use these browsers combined [16]. For the purposes of our testing, we have our controller only apply the proposed request rerouting mechanism to h1 - this allows us to compare with h4, that will only be using the standard DNS request rerouting mechanism. It should be important to note that both of these browsers perform link prefetching and DNS caching of at least one minute. The end users are configured to use h5 as their primary DNS server. End user automation is achieved via a series of javascript commands embedded in thesis.com, which causes the end user to automatically forward through pages at set intervals and after certain events, like after a page completely loads. On both chrome and firefox, the cache has been disabled (But not the DNS cache) in order to force the web browsers to fully download thesis.com every time they visit it. Other tools such as AutoHotKey and shell scripting is used to further automate the timing, execution, and recording of the testing procedures. 30

33 3.3.2 Testing procedures This subsection describes the testing procedures we perform, and the methodology and reasoning behind them. We also outline the goals of our testing procedures. To recap, these are the design goals we aim to achieve with our request rerouting mechanism: 1. Able to work around DNS caching/prefetching 2. Able to fully utilise network information 3. Easy to configure 4. Incrementally implementable 5. Able to utilise and act on the ability to rapidly create virtual CDN replica servers 6. Scalable for network and service expansion 7. Able to provide access to the CDN for end users that are not customers item of the ISP From these, the ones we can directly test are goals 1, 2, and 5. The testing procedures we thus create must therefore be able to not only test for goals 1, 2, and 5, but also ideally test for the system performance in achieving those goals 10. To that end, the following test procedures were created: Request rerouting after link failure Request rerouting during link congestion Request rerouting with the introduction of a new CDN replica server Also, within each of these procedures, we define two types of tests: Passive user and Active user. Because the ultimate goal of a CDN is to increase QoE and QoS for their end users, it is therefore desirable to model expected user behaviour in times of duress. As such, we model a scenario where the user tries to access content via simply typing the URL into their browser, and another scenario where if congestion is experienced, or load time is greater than 20 seconds, the user refreshes the page Request rerouting after link failure In this test, we observe the effectiveness of our proposed request rerouting mechanism after a CDN replica server experiences a link failure. Here, we are primarily concerned with content access delay. First, the end user visits thesis.com, which points to CDN replica server h2. Doing so causes them to perform a DNS request for h2. They are assumed to stay on the landing 10 bitrate, content access delay, etc. 31

34 page of thesis.com for 10 seconds while they read some of the news. During this time period, a shell script is called that disables the link between h2 and s1, and updates the controller via the REST API that h2 has gone down, and that requests for h2 should be redirected towards h3. At the same time, the DNS server at h5 updates it s A record to reflect that fact. After the end user has stayed on the landing page of thesis.com for 10 seconds, they then click on a hyperlink to thesis.com/firstpage.html, which contains more content. We then time how long it takes for that page to successfully load. In the passive user case, if the page takes longer than one minute to load, we end the test. In the active user case, the user attempts to refresh the page if there are no signs of activity after 20 seconds Request rerouting during link congestion In this test, we observe the effectiveness of our proposed request rerouting mechanism when our CDN replica server is under heavy congestion. In this scenario, we simulate the link between s1 and h2 (The switch and a CDN replica server) as being congested. This is achieved by lowering the bandwidth on the link between s1 and h2 to 1Mbps. As with the previous test, thesis.com initially points to h2. The test begins when a user visits thesis.com. The user stays on thesis.com for 10 seconds, and is then redirected to thesis.com/firstpage.html. During this 10 second period, a shell script is executed that notifies the controller(c0) that CDN replica h3 is available, and uncongested. Correspondingly, the DNS server at h5 updates it s A record for thesis.com to point to CDN replica h3. We then time how long it takes for the user to complete a full trip around the thesis.com testpages. After a full trip, when the user is back on the thesis.com landing page, we then time how much longer it takes to for the user to connect to the new CDN replica server(h3) if they have not already done so Request rerouting with the introduction of a new CDN replica server This test can be seen as an extension of the previous test. The purpose of this test is to examine the impact that the proposed request rerouting mechanism might have on users who are not located in the ISP s network, but yet are still using the CDN service 11. This test will also provide insight on how rapid establishment of a new CDN server depending on server demand can have an effect on both the ISP s customers, but also visitors to the ISP s CDN network. In this scenario, both h1 and h4 (an end user in the ISP network, and an end user not in the ISP network) are initially connected to h2. We make the link between h2 and s1 constrained to 1Mbps, so as to easier simulate the effects of congestion. As with previous tests, both hosts visit the landing page of thesis.com initially, and then wait for 10 seconds. They then visit thesis.com/firstpage.html, causing the link between s1 and h2 to become congested. At the same time that this link becomes congested, a 30 second timer is started. At the end of this 30 second timer, h3 is brought online, and the controller is notified that there is now a new, more optimal CDN replica server available. Throughout these tests, we are timing how long it takes for h1 and h4 to reach 11 For example, they may be logging into a content portal while at work 32

35 the initial thesis.com landing page again, compared to the case where we don t use this thesis proposed method of request rerouting. 33

36 Chapter 4 Results & Performance This chapter contains the results and performance of our proposed request rerouting mechanism. Observations relating to the testing process are also made. 4.1 Request rerouting after link failure The results of 50 tests are shown in Table 4.1 below: Firefox Chrome mean std mean std Reference mean load time With improved request rerouting mechanism Without improved request rerouting mechanism; active user 75.3 or or Without improved request rerouting mechanism; passive user N/A N/A N/A N/A Table 4.1: Time taken to fully traverse and load the thesis.com testpages, in seconds. The reference time is the time taken to load the website directly from the CDN replica server, without any form of redirection. With both Firefox and Chrome, when the browsers were left alone as described in passive testing mode, both failed to load a page after over a minute, and so are both counted as timeouts. Interesting behaviour was observed for the active testing mode in both Firefox and chrome. In Firefox, two different, but locally similar sets of results were gathered; One with a mean total load time of 75.3 seconds, and another with a mean total load time of seconds, a 40 second difference. This was caused by two different behaviours during active testing: A refresh was performed after 20 seconds of inactivity, as outlined in Subsection Upon receiving no result, another refresh was done at 25 seconds, 30 seconds, etc. until the website finally started loading. Activity was consistently observed at 100 seconds after thesis.com was initially loaded. This behaviour corresponds to the mean load time. 34

37 A refresh was performed after 50 seconds of inactivity (60 seconds total since the start of the testing procedure). Activity was consistently observed immediately after this reload. This behaviour corresponds to the 75.3 mean load time. As shown in the table, the average mean load time for Chrome is at least 30 seconds higher than that for firefox. The active behaviour was similar - refresh at 20 seconds, then 25, 30, etc. until activity is observed. In this case, the page access delay was a minute; when the page was refreshed after a total of a minute s delay since the browser visited thesis.com, the first testpage appeared to start loading. However, the page didn t fully load - Only seven images out of the ten would load on firstpage.html. After 20 seconds, the page was refreshed and the page loaded fine; but the same behaviour occured on secondpage.html, as well. This behaviour appeared consistently, and contributed majorly to the extra time observed in Table 4.1. The results from this testing procedure indicate that our proposed request rerouting system can successfully reroute end users after a link failure. However, there is an observable performance cost, with mean load time taking around 5 seconds more with our proposed mechanism compared to the reference. However, after the 10 second delay on thesis.com, the access delay until firstpage.html started loading was negligible. 4.2 Request rerouting during link congestion The results of 50 tests are shown in Table 4.2 below: Firefox Chrome mean std mean std Reference mean load time(fast) Reference mean load time(congested) Load time with improved request rerouting mechanism Time until connected to new CDN server, with improved request rerouting mechanism Load time without improved request rerouting mechanism Time until connected to new CDN server, without improved request rerouting mechanism Table 4.2: Time taken to fully traverse and load the thesis.com testpages, in seconds. Also shown is time delay until the client connects to the optimal CDN server over the congested one. There are two reference times in this test. The fast one is identical to the one presented in the previous test. The slow one is a reference for the amount of time it takes to fully load thesis.com and it s testpages across the congested link with 0 request redirection. In the tests without the improved request rerouting mechanism, both browsers beat the congested reference mean load time. This is because, between the second and third testpage, they would change connections from the congested h2, to the uncongested h3. However, as shown in the table, there is a large discrepancy between Chrome and Firefox 35

38 in the delay until the uncongested h3 server was connected to. Chrome would connect to h3 half-way through loading the second testpage - one of the GET requests performed while loading the second test page would be sent to h3, instead of h2. For Firefox however, h3 would only be connected to once the second testpage had finished loading - Firefox would only connect to h3 between page loads. This difference did not cause much of a discrepancy between the mean load time for the entire set of testpages however, as the Chrome browser only had a chance to load one extra image from the uncongested h3 server over the firefox browser. As with the previous test, our improved request rerouting mechanism shows that it is capable of providing request rerouting in a way that utilises current network information. Similarly also, the mean load time is around 5 seconds higher than the reference time. Users that are successfully rerouted by our rerouting system must have their first few TCP packets pass into the controller, in order to determine if the CDN node they originally are destined for is the one that is optimal for them. This handling and processing of packets by the controller may be what is causing that 5 second gap. 4.3 Request rerouting with the introduction of a new CDN replica server The results of 50 tests are shown in Table 4.3 below: h1 h4 Reference mean load time (together) mean: 236 std: 2.26 Mean load time with new CDN introduced mean: 56.9 std: 17.4 mean:135.2 std: 16.7 Table 4.3: Time taken to fully traverse and load the thesis.com testpages, in seconds. The singular reference load time in this test indicates how long it took on average for both hosts, h1 and h4 to finish loading the thesis.com testpages when both hosts were accessing the congested CDN replica server h2 simultaneously. It is quite surprising to see that the standard deviation is comparatively low for this measurement. When this measurement was being taken, the two hosts would appear to load the website in random chunks, and take turns in downloading content from h2. This measurement measures the entire time that h2 is put under load, so even if one host has finished the thesis.com testpages, the measurement continues until the other hosts finishes their run. The large values for standard deviations seen in the other measurements here are caused via h1. If h1 was occupied with GET request responses from h2 at the 30 second mark, then it could take ten to twenty seconds for h1 to finish those downloads, and then get rerouted by our proposed mechanism. This had a flow on effect onto h4, whereby it would sometimes have to wait for h1 to use another link, so h4 could utilise the now uncongested link. As can be seen, the mean time for h1 to finish loading the thesis.com testpages is less than half that of h4 s. Similarly, mean time for h4 to finish loading the thesis.com testpages is almost half that of the reference time. This indicates that our proposed 36

39 ISP CDN request rerouting system can be used to utilise the introduction of rapidly established CDN replica servers to increase QoE for both users of the ISP s network, but also for users outside the ISP s network by indirectly reducing the load on their links. 37

40 Chapter 5 Evaluation For the sake of convenience, we will reiterate the design goals specified in Chapter 3 here. Our proposed design should be: 1. Able to work around DNS caching/prefetching 2. Able to fully utilise network information 3. Easy to configure 4. Incrementally implementable 5. Able to utilise and act on the ability to rapidly create virtual CDN replica servers 6. Scalable for network and service expansion 7. Able to provide access to the CDN for end users that are not customers item of the ISP We will now assess the effectiveness of our proposed request rerouting mechanism for ISP operated CDNs by discussing and evaluating the results achieved in the previous chapter. 5.1 Results The experiments outlined in Section were designed specifically to test if our design met goals 1, 2, and 5. As the results in Chapter 4 indicate, our proposed implementation of an ISP CDN s request rerouting mechanism via NAT augmented DNs is successful in those areas. The simplicity of the table-miss flow rules employed by our Ryu controller as defined in Section , and the usage of simple shell scripts and REST APIs indicate that our system should be easy to integrate and configure in an ISP s network, thus fulfilling goal 3. Because the proposed request rerouting mechanism is to be used as an addition to traditional DNS routing mechanisms, it is possible for an ISP to implement that first, and then incrementally introduce OpenFlow-enabled switches throughout their network as demand requires. This fulfills goal 4. Goal 7 is similarly fulfilled: Users outside of the ISP network can simply use the traditional DNS request rerouting method, they 38

41 just wont get the benefits as demonstrated in this thesis. The only goal not covered by the results of our experiments and/or mechanism design is goal 6; whether or not our proposed request redirection mechanism is scalable. The results of our testing also confirms our claims about the ill effects of client side DNS caching. For example, in our first test procedure, browsers simply timed out, or had users refreshing for over a minute in order to get any kind of response. 5.2 Discussion Our proposed request rerouting mechanism for ISP operated CDNs ended up being quite visibly successful. Load times were often cut in half 1 or more, compared to a request rerouting system such as traditional DNS. Granted, the situations our tests described were edge cases that don t happen very often in current enterprise networks. But as Akamai and Justine Sherry et al. [3, 37] claim, the way forward for ISPs and CDNs in the future involves the usage of cloud services to rapidly adapt to changes in the networking landscape and demands. This means that situations where our proposed request rerouting mechanisms are beneficial will only increase in the future. The results of the thorough investigation into the effects of client side DNS caching are a bit surprising. Issues like how timing the refreshing of a page has an effect on how soon the page successfully refreshes are still not fully understood by the author. While we certainly expected some errors to occur due to DNS caching such as pages timing out, timing errors certainly were not. Unexpected behaviour such as this is exactly why such an investigation had to be undertaken, however. Unfortunately, a rigorous examination on the scalability of of our proposed mechanism has not been undertaken. Because of the current infancy of SDN technology compared to legacy networks, published performance and scalability research is not yet widely available, especially for a purely reactive 2 system like the one we have proposed. Indeed, this may be the cause of the 5 second gap between the reference timing and the successfully routed timing in our first two networking experiments. Adam Zarek [40] has done some research into the impact of flow timeout length on SDN performance, and indicates that a heuristic matching on TCP patterns is desirable. Given that the flow rules added by our controller are all TCP, and are designed to be short lived 3, future work on determining the scalability of our proposed mechanism looks promising. A potential solution to possible future scalability problems is to move our proposed request rerouting mechanism away from the ISP s edge router, and instead into the home routers and devices of home owners. This way, instead of a singular device having to 1 Note that this depends on network context; rerouting someone from a 10Mbit connection to a 5Mbit connection will obviously increase their load time. In the context of CDN request rerouting however, the reason you re performing rerouting in the first place is to provide them with a faster connection to content. 2 A reactive SDN system is one in which all new flows are sent to the controller for inspection. In our set up, all previously unseen TCP flows are sent to the controller to determine if it is going to the right CDN node or not. 3 Our flow rules are only of use when there s a possibility of a user utilising cached DNS information, or if we want to temporarily balance network congestion. 39

42 handle thousands of flows, a single device handles tens of flows per household. Similar networking tools and APIs have already been researched [41], which indicates that future work on this request rerouting tool could indeed be brought into the home networking space. 40

43 Chapter 6 Conclusion In this thesis we have successfully demonstrated a request rerouting mechanism for ISP operated CDNs. It is able to transparently reroute users away from nonoptimal CDN replica servers to optimal ones in the situation that the user has stale DNS information due to client side DNS caching, or if network conditions require it. It utilises SDN techniques to achieve this, using the OpenFlow1.3 spec on an OpenFlow enabled switch running in a mininet virtualised network. A controller operating on the Ryu framework allows the request rerouting mechanism to occur, by installing NAT and DNAT rules that act on TCP flows. Such a request rerouting mechanism can find much use in both ISP operated CDNs, as well as ISPs that are collaborating with CDNs to provide a unified service. Current research shows that future requirements and design goals for ISPs include the ability to rapidly implement cloud-based services. Our proposed request rerouting mechanism offers an easily configured, incrementally implementable solution that can perform that role exactly. We have also performed a first of it s kind analysis of the effects of client side DNS caching on DNS based CDN request rerouting mechanisms. This analysis revealed that client side DNS caching can lead to a degradation in Quality of Experience and Quality of Service experienced by the end user performing that caching. 6.1 Future Work Because this thesis dealt mainly with the request rerouting mechanism, and not the request rerouting algorithm to go with it, the natural extension of this work is to develop a request rerouting algorithm that can utilise the request rerouting mechanism proposed by this thesis efficiently. Research could also be done on the scalability of the request rerouting mechanism outlined in this thesis, as tests have so far only been performed on a virtualised network on a desktop PC. 41

44 Bibliography [1] J. Brodkin, Its not a fast lane but comcast built a cdn to charge for video delivery. Arstechnica, accessed 09/2014, [2] The telstra media blueprint. White paper, Telstra, [3] B. Frank et al., Pushing cdn-isp collaboration to the limit, ACM SIGCOMM Computer Communication Review, vol. 43, no. 3, pp , [4] J. S. Otto, M. A. Sánchez, J. P. Rula, and F. E. Bustamante, Content delivery and the natural evolution of dns: Remote dns trends, performance issues and alternative solutions, in Proceedings of the 2012 ACM Conference on Internet Measurement Conference, IMC 12, (New York, NY, USA), pp , ACM, [5] E. Nygren, R. K. Sitaraman, and J. Sun, The akamai network: A platform for highperformance internet applications, SIGOPS Oper. Syst. Rev., vol. 44, pp. 2 19, Aug [6] B. Ager, N. Chatzis, A. Feldmann, N. Sarrar, S. Uhlig, and W. Willinger, Anatomy of a large european ixp, in Proceedings of the ACM SIGCOMM 2012 Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, SIGCOMM 12, (New York, NY, USA), pp , ACM, [7] The content delivery network (cdn): Deliverying the ultimate web experience. White paper, Bell Canada, [8] Cisco visual networking index: Forecast and methodology, White paper, Cisco, [9] A. G. J. But Jason, Keller Urs, Title a rationale for web caching in consumer isps : the impact of dns lookup times and http session characteristics. Swinburne University of Technology, [10] A. GRAVEY, F. GUILLEMIN, and S. MOTEAU, Last Mile Caching of Video Content by an ISP, in ETS 2013 : 2nd European Teletraffic Seminar, no , (Karlskrona, Sweden), p.., Sept

45 [11] Netflix open connect, a content delivery networkprovided by netflix. Information accessible at: Last accessed October [12] A. Pathan and R. Buyya, A taxonomy and survey of content delivery networks. Technical Report, GRIDS-TR , Grid Computing and Distributed Systems Laboratory, The University of Melbourne, Australia., Feb [13] L. Wang, V. Pai, and L. Peterson, The effectiveness of request redirection on cdn robustness, SIGOPS Oper. Syst. Rev., vol. 36, pp , Dec [14] B. Cain, A. Barbir, R. Nair, and O. Spatscheck, Known content network (cn) request-routing mechanisms. RFC 5685, [15] Load balancing without load balancers. Industry blog, CloudFlare, Accessible at: [16] Browser statistics. Browser statistics, w3schools, Accessible at: stats.asp. [17] Dns ttl not honored. Chromium issues board, Google, Accessible at: The final comment explains chrome s current DNS cache behaviour. [18] Network.dnscacheexpiration - mozillazine knowledge base. Mozilla product documentation, Mozilla, Accessible at: [19] How internet explorer uses the cache for dns host entries. Microsoft support documentation, Microsoft. Accessible at: [20] P. Smith, Professional website performance: Optimizing the Front-End and Back- End. John Wiley & Sons, [21] Dns prefetching - the chromium projects. Chromium design documents, Google. Accesible at: [22] Link prefetching faq. Mozilla design documents, Mozilla. Accessible at: prefetching FAQ. [23] Getting to the content you want, faster in ie11. Internet Explorer design blog, Microsoft, Accessible at: blogs.msdn.com/b/ie/archive/2013/12/04/gettingto-the-content-you-want-faster-in-ie11.aspx?redirected=true. [24] How anycast routing is used at maxcdn. Industry blog, MaxCDN, Accessible at: [25] M. Prince, A brief primer on anycast. Industry blog, Cloudflare, Accessible at: 43

46 [26] Openflow: enabling innovation in campus networks. White paper, OpenFlow, [27] Ryu SDN controller framework, main site accessible at: [28] Cdni data tracker. Datatracker page for CDNI, IETF. Accessible at: [29] D. C. T. K. M-k Shin, H-J Kim, Cdni request routing with sdn. IETF draft, [30] Y. E. J. Y. C. D. Jingguo Ge, Yulei Wu, Investigation of Anycast implementation in Software-Defined Networking. CRC Press, [31] P. Georgopoulos, M. Broadbent, B. Plattner, and N. Race, Cache as a service. IEEE, [32] J. V. Doorn, Building a large scale cdn with apache traffic server. Presented at ApacheCon North America, [33] J. Graham-Cumming, Route leak incident on october 2, Industry blog, Cloudflare, Accessible at: [34] H. Ballani and P. Francis, Understanding ip anycast, tech. rep., Cornell University, [35] F. Dobrian, V. Sekar, A. Awan, I. Stoica, D. Joseph, A. Ganjam, J. Zhan, and H. Zhang, Understanding the impact of video quality on user engagement, ACM SIGCOMM Computer Communication Review, vol. 41, no. 4, pp , [36] M. Luhar, game of thrones premiere crashes hbo go; 2nd big outage. Online news article, scpr.org, Accessible at: [37] J. Sherry, S. Hasan, C. Scott, A. Krishnamurthy, S. Ratnasamy, and V. Sekar, Making middleboxes someone else s problem: network processing as a cloud service, ACM SIGCOMM Computer Communication Review, vol. 42, no. 4, pp , [38] Mininet Network virtualisation tool, main site accessible at: [39] BIND9, a DNS Nameserver utility, main site accessible at: [40] A. Zarek, Openflow timeouts demystified, Master s thesis, University of Toronto, Ontario, Canada, [41] H. Gharakheili, J. Bass, L. Exton, and V. Sivaraman, Personalizing the home network experience using cloud-based sdn, in A World of Wireless, Mobile and Multimedia Networks (WoWMoM), 2014 IEEE 15th International Symposium on, pp. 1 6, June

47 Appendix 1.1 An example of a DNS request for facebook.com, a CDN hosted website. Figure 1: A wireshark trace of a DNS request for facebook.com.2 HTML for wikipedia.com <!DOCTYPE HTML PUBLIC //IETF//DTD HTML 2. 0//EN > <html><head> <t i t l e >301 Moved Permanently</ t i t l e > </head><body> <h1>moved Permanently </h1> <p>the document has moved <a h r e f= http : / /www. wikipedia. org/ >here </a>.</p> </body></html>.3 The Ryu controller for my thesis. import j s o n import l o g g i n g 45

Rapid IP redirection with SDN and NFV. Jeffrey Lai, Qiang Fu, Tim Moors December 9, 2015

Rapid IP redirection with SDN and NFV. Jeffrey Lai, Qiang Fu, Tim Moors December 9, 2015 Rapid IP redirection with SDN and NFV Jeffrey Lai, Qiang Fu, Tim Moors December 9, 2015 Background Enabling ISP-CDN collaboration SDN, NFV, CDN Basics Client assumptions And breaking them The problem My

More information

Request Routing, Load-Balancing and Fault- Tolerance Solution - MediaDNS

Request Routing, Load-Balancing and Fault- Tolerance Solution - MediaDNS White paper Request Routing, Load-Balancing and Fault- Tolerance Solution - MediaDNS June 2001 Response in Global Environment Simply by connecting to the Internet, local businesses transform themselves

More information

Web Caching and CDNs. Aditya Akella

Web Caching and CDNs. Aditya Akella Web Caching and CDNs Aditya Akella 1 Where can bottlenecks occur? First mile: client to its ISPs Last mile: server to its ISP Server: compute/memory limitations ISP interconnections/peerings: congestion

More information

FortiBalancer: Global Server Load Balancing WHITE PAPER

FortiBalancer: Global Server Load Balancing WHITE PAPER FortiBalancer: Global Server Load Balancing WHITE PAPER FORTINET FortiBalancer: Global Server Load Balancing PAGE 2 Introduction Scalability, high availability and performance are critical to the success

More information

Global Server Load Balancing

Global Server Load Balancing White Paper Overview Many enterprises attempt to scale Web and network capacity by deploying additional servers and increased infrastructure at a single location, but centralized architectures are subject

More information

A Link Load Balancing Solution for Multi-Homed Networks

A Link Load Balancing Solution for Multi-Homed Networks A Link Load Balancing Solution for Multi-Homed Networks Overview An increasing number of enterprises are using the Internet for delivering mission-critical content and applications. By maintaining only

More information

APPLICATION-AWARE ROUTING IN SOFTWARE-DEFINED NETWORKS

APPLICATION-AWARE ROUTING IN SOFTWARE-DEFINED NETWORKS APPLICATION-AWARE ROUTING IN SOFTWARE-DEFINED NETWORKS SARO VELRAJAN Director of Technology, Aricent APPLICATION-AWARE ROUTING IN SOFTWARE-DEFINED NETWORKS Service providers are constantly exploring avenues

More information

Distributed Systems. 23. Content Delivery Networks (CDN) Paul Krzyzanowski. Rutgers University. Fall 2015

Distributed Systems. 23. Content Delivery Networks (CDN) Paul Krzyzanowski. Rutgers University. Fall 2015 Distributed Systems 23. Content Delivery Networks (CDN) Paul Krzyzanowski Rutgers University Fall 2015 November 17, 2015 2014-2015 Paul Krzyzanowski 1 Motivation Serving web content from one location presents

More information

Distributed Systems. 25. Content Delivery Networks (CDN) 2014 Paul Krzyzanowski. Rutgers University. Fall 2014

Distributed Systems. 25. Content Delivery Networks (CDN) 2014 Paul Krzyzanowski. Rutgers University. Fall 2014 Distributed Systems 25. Content Delivery Networks (CDN) Paul Krzyzanowski Rutgers University Fall 2014 November 16, 2014 2014 Paul Krzyzanowski 1 Motivation Serving web content from one location presents

More information

Superior Disaster Recovery with Radware s Global Server Load Balancing (GSLB) Solution

Superior Disaster Recovery with Radware s Global Server Load Balancing (GSLB) Solution Superior Disaster Recovery with Radware s Global Server Load Balancing (GSLB) Solution White Paper January 2012 Radware GSLB Solution White Paper Page 1 Table of Contents 1. EXECUTIVE SUMMARY... 3 2. GLOBAL

More information

AKAMAI WHITE PAPER. Delivering Dynamic Web Content in Cloud Computing Applications: HTTP resource download performance modelling

AKAMAI WHITE PAPER. Delivering Dynamic Web Content in Cloud Computing Applications: HTTP resource download performance modelling AKAMAI WHITE PAPER Delivering Dynamic Web Content in Cloud Computing Applications: HTTP resource download performance modelling Delivering Dynamic Web Content in Cloud Computing Applications 1 Overview

More information

Internet Protocol: IP packet headers. vendredi 18 octobre 13

Internet Protocol: IP packet headers. vendredi 18 octobre 13 Internet Protocol: IP packet headers 1 IPv4 header V L TOS Total Length Identification F Frag TTL Proto Checksum Options Source address Destination address Data (payload) Padding V: Version (IPv4 ; IPv6)

More information

Distributed Systems 19. Content Delivery Networks (CDN) Paul Krzyzanowski pxk@cs.rutgers.edu

Distributed Systems 19. Content Delivery Networks (CDN) Paul Krzyzanowski pxk@cs.rutgers.edu Distributed Systems 19. Content Delivery Networks (CDN) Paul Krzyzanowski pxk@cs.rutgers.edu 1 Motivation Serving web content from one location presents problems Scalability Reliability Performance Flash

More information

GLOBAL SERVER LOAD BALANCING WITH SERVERIRON

GLOBAL SERVER LOAD BALANCING WITH SERVERIRON APPLICATION NOTE GLOBAL SERVER LOAD BALANCING WITH SERVERIRON Growing Global Simply by connecting to the Internet, local businesses transform themselves into global ebusiness enterprises that span the

More information

The OpenDNS Global Network Delivers a Secure Connection Every Time. Everywhere.

The OpenDNS Global Network Delivers a Secure Connection Every Time. Everywhere. The OpenDNS Global Network Delivers a Secure Connection Every Time. Everywhere. Network Performance Users devices create multiple simultaneous connections each time we exchange data with other Internet

More information

Intelligent Content Delivery Network (CDN) The New Generation of High-Quality Network

Intelligent Content Delivery Network (CDN) The New Generation of High-Quality Network White paper Intelligent Content Delivery Network (CDN) The New Generation of High-Quality Network July 2001 Executive Summary Rich media content like audio and video streaming over the Internet is becoming

More information

How To Understand The Power Of A Content Delivery Network (Cdn)

How To Understand The Power Of A Content Delivery Network (Cdn) Overview 5-44 5-44 Computer Networking 5-64 Lecture 8: Delivering Content Content Delivery Networks Peter Steenkiste Fall 04 www.cs.cmu.edu/~prs/5-44-f4 Web Consistent hashing Peer-to-peer CDN Motivation

More information

Measuring the Web: Part I - - Content Delivery Networks. Prof. Anja Feldmann, Ph.D. Dr. Ramin Khalili Georgios Smaragdakis, PhD

Measuring the Web: Part I - - Content Delivery Networks. Prof. Anja Feldmann, Ph.D. Dr. Ramin Khalili Georgios Smaragdakis, PhD Measuring the Web: Part I - - Content Delivery Networks Prof. Anja Feldmann, Ph.D. Dr. Ramin Khalili Georgios Smaragdakis, PhD Acknowledgement Material presented in these slides is borrowed from presentajons

More information

Internet Firewall CSIS 4222. Packet Filtering. Internet Firewall. Examples. Spring 2011 CSIS 4222. net15 1. Routers can implement packet filtering

Internet Firewall CSIS 4222. Packet Filtering. Internet Firewall. Examples. Spring 2011 CSIS 4222. net15 1. Routers can implement packet filtering Internet Firewall CSIS 4222 A combination of hardware and software that isolates an organization s internal network from the Internet at large Ch 27: Internet Routing Ch 30: Packet filtering & firewalls

More information

Teridion. Rethinking Network Performance. The Internet. Lightning Fast. Technical White Paper July, 2015 www.teridion.com

Teridion. Rethinking Network Performance. The Internet. Lightning Fast. Technical White Paper July, 2015 www.teridion.com Teridion The Internet. Lightning Fast. Rethinking Network Performance Technical White Paper July, 2015 www.teridion.com Executive summary Online services face the growing dual challenge of supporting many

More information

Content Delivery Networks

Content Delivery Networks Content Delivery Networks Terena 2000 ftp://ftpeng.cisco.com/sgai/t2000cdn.pdf Silvano Gai Cisco Systems, USA Politecnico di Torino, IT sgai@cisco.com Terena 2000 1 Agenda What are Content Delivery Networks?

More information

Web Application Hosting Cloud Architecture

Web Application Hosting Cloud Architecture Web Application Hosting Cloud Architecture Executive Overview This paper describes vendor neutral best practices for hosting web applications using cloud computing. The architectural elements described

More information

Distributed Systems. 24. Content Delivery Networks (CDN) 2013 Paul Krzyzanowski. Rutgers University. Fall 2013

Distributed Systems. 24. Content Delivery Networks (CDN) 2013 Paul Krzyzanowski. Rutgers University. Fall 2013 Distributed Systems 24. Content Delivery Networks (CDN) Paul Krzyzanowski Rutgers University Fall 2013 November 27, 2013 2013 Paul Krzyzanowski 1 Motivation Serving web content from one location presents

More information

The Requirement for a New Type of Cloud Based CDN

The Requirement for a New Type of Cloud Based CDN The Requirement for a New Type of Cloud Based CDN Executive Summary The growing use of SaaS-based applications has highlighted some of the fundamental weaknesses of the Internet that significantly impact

More information

Flexible SDN Transport Networks With Optical Circuit Switching

Flexible SDN Transport Networks With Optical Circuit Switching Flexible SDN Transport Networks With Optical Circuit Switching Multi-Layer, Multi-Vendor, Multi-Domain SDN Transport Optimization SDN AT LIGHT SPEED TM 2015 CALIENT Technologies 1 INTRODUCTION The economic

More information

WAN Traffic Management with PowerLink Pro100

WAN Traffic Management with PowerLink Pro100 Whitepaper WAN Traffic Management with PowerLink Pro100 Overview In today s Internet marketplace, optimizing online presence is crucial for business success. Wan/ISP link failover and traffic management

More information

THE MASTER LIST OF DNS TERMINOLOGY. v 2.0

THE MASTER LIST OF DNS TERMINOLOGY. v 2.0 THE MASTER LIST OF DNS TERMINOLOGY v 2.0 DNS can be hard to understand and if you re unfamiliar with the terminology, learning more about DNS can seem as daunting as learning a new language. To help people

More information

Chapter 5. Data Communication And Internet Technology

Chapter 5. Data Communication And Internet Technology Chapter 5 Data Communication And Internet Technology Purpose Understand the fundamental networking concepts Agenda Network Concepts Communication Protocol TCP/IP-OSI Architecture Network Types LAN WAN

More information

LOAD BALANCING IN WEB SERVER

LOAD BALANCING IN WEB SERVER LOAD BALANCING IN WEB SERVER Renu Tyagi 1, Shaily Chaudhary 2, Sweta Payala 3 UG, 1,2,3 Department of Information & Technology, Raj Kumar Goel Institute of Technology for Women, Gautam Buddh Technical

More information

OpenFlow Based Load Balancing

OpenFlow Based Load Balancing OpenFlow Based Load Balancing Hardeep Uppal and Dane Brandon University of Washington CSE561: Networking Project Report Abstract: In today s high-traffic internet, it is often desirable to have multiple

More information

SOFTWARE-DEFINED NETWORKING AND OPENFLOW

SOFTWARE-DEFINED NETWORKING AND OPENFLOW SOFTWARE-DEFINED NETWORKING AND OPENFLOW Freddie Örnebjär TREX Workshop 2012 2012 Brocade Communications Systems, Inc. 2012/09/14 Software-Defined Networking (SDN): Fundamental Control

More information

Alteon Global Server Load Balancing

Alteon Global Server Load Balancing Alteon Global Server Load Balancing Whitepaper GSLB Operation Overview Major Components Distributed Site Monitoring Distributed Site State Protocol Internet Topology Awareness DNS Authoritative Name Server

More information

Evolution of OpenCache: an OpenSource Virtual Content Distribution Network (vcdn) Platform

Evolution of OpenCache: an OpenSource Virtual Content Distribution Network (vcdn) Platform Evolution of OpenCache: an OpenSource Virtual Content Distribution Network (vcdn) Platform Daniel King d.king@lancaster.ac.uk Matthew Broadbent m.broadbent@lancaster.ac.uk David Hutchison d.hutchison@lancaster.ac.uk

More information

Traffic delivery evolution in the Internet ENOG 4 Moscow 23 rd October 2012

Traffic delivery evolution in the Internet ENOG 4 Moscow 23 rd October 2012 Traffic delivery evolution in the Internet ENOG 4 Moscow 23 rd October 2012 January 29th, 2008 Christian Kaufmann Director Network Architecture Akamai Technologies, Inc. way-back machine Web 1998 way-back

More information

THE MASTER LIST OF DNS TERMINOLOGY. First Edition

THE MASTER LIST OF DNS TERMINOLOGY. First Edition THE MASTER LIST OF DNS TERMINOLOGY First Edition DNS can be hard to understand and if you re unfamiliar with the terminology, learning more about DNS can seem as daunting as learning a new language. To

More information

DATA COMMUNICATOIN NETWORKING

DATA COMMUNICATOIN NETWORKING DATA COMMUNICATOIN NETWORKING Instructor: Ouldooz Baghban Karimi Course Book: Computer Networking, A Top-Down Approach, Kurose, Ross Slides: - Course book Slides - Slides from Princeton University COS461

More information

KASPERSKY DDOS PROTECTION. Discover how Kaspersky Lab defends businesses against DDoS attacks

KASPERSKY DDOS PROTECTION. Discover how Kaspersky Lab defends businesses against DDoS attacks KASPERSKY DDOS PROTECTION Discover how Kaspersky Lab defends businesses against DDoS attacks CYBERCRIMINALS ARE TARGETING BUSINESSES If your business has ever suffered a Distributed Denial of Service (DDoS)

More information

TRUFFLE Broadband Bonding Network Appliance. A Frequently Asked Question on. Link Bonding vs. Load Balancing

TRUFFLE Broadband Bonding Network Appliance. A Frequently Asked Question on. Link Bonding vs. Load Balancing TRUFFLE Broadband Bonding Network Appliance A Frequently Asked Question on Link Bonding vs. Load Balancing 5703 Oberlin Dr Suite 208 San Diego, CA 92121 P:888.842.1231 F: 858.452.1035 info@mushroomnetworks.com

More information

Content Delivery Networks (CDN) Dr. Yingwu Zhu

Content Delivery Networks (CDN) Dr. Yingwu Zhu Content Delivery Networks (CDN) Dr. Yingwu Zhu Web Cache Architecure Local ISP cache cdn Reverse Reverse Proxy Reverse Proxy Reverse Proxy Proxy L4 Switch Content Content Content Server Content Server

More information

Data Center Content Delivery Network

Data Center Content Delivery Network BM 465E Distributed Systems Lecture 4 Networking (cont.) Mehmet Demirci Today Overlay networks Data centers Content delivery networks Overlay Network A virtual network built on top of another network Overlay

More information

Service Definition. Internet Service. Introduction. Product Overview. Service Specification

Service Definition. Internet Service. Introduction. Product Overview. Service Specification Service Definition Introduction This Service Definition describes Nexium s from the customer s perspective. In this document the product is described in terms of an overview, service specification, service

More information

SiteCelerate white paper

SiteCelerate white paper SiteCelerate white paper Arahe Solutions SITECELERATE OVERVIEW As enterprises increases their investment in Web applications, Portal and websites and as usage of these applications increase, performance

More information

Microsoft Office Communications Server 2007 R2

Microsoft Office Communications Server 2007 R2 Microsoft Office Communications Server 2007 R2 Scale to a Load Balanced Enterprise Edition Pool with WebMux Walkthrough Published: Sept. 2009 For the most up-to-date version of the Scale to a Load Balanced

More information

co Characterizing and Tracing Packet Floods Using Cisco R

co Characterizing and Tracing Packet Floods Using Cisco R co Characterizing and Tracing Packet Floods Using Cisco R Table of Contents Characterizing and Tracing Packet Floods Using Cisco Routers...1 Introduction...1 Before You Begin...1 Conventions...1 Prerequisites...1

More information

HUAWEI OceanStor 9000. Load Balancing Technical White Paper. Issue 01. Date 2014-06-20 HUAWEI TECHNOLOGIES CO., LTD.

HUAWEI OceanStor 9000. Load Balancing Technical White Paper. Issue 01. Date 2014-06-20 HUAWEI TECHNOLOGIES CO., LTD. HUAWEI OceanStor 9000 Load Balancing Technical Issue 01 Date 2014-06-20 HUAWEI TECHNOLOGIES CO., LTD. Copyright Huawei Technologies Co., Ltd. 2014. All rights reserved. No part of this document may be

More information

Measuring CDN Performance. Hooman Beheshti, VP Technology

Measuring CDN Performance. Hooman Beheshti, VP Technology Measuring CDN Performance Hooman Beheshti, VP Technology Why this matters Performance is one of the main reasons we use a CDN Seems easy to measure, but isn t Performance is an easy way to comparison shop

More information

Demand Routing in Network Layer for Load Balancing in Content Delivery Networks

Demand Routing in Network Layer for Load Balancing in Content Delivery Networks Demand Routing in Network Layer for Load Balancing in Content Delivery Networks # A SHRAVANI, 1 M.Tech, Computer Science Engineering E mail: sravaniathome@gmail.com # SYED ABDUL MOEED 2 Asst.Professor,

More information

CDN INTEROPERABILITY DEFINING STANDARDS THAT CAN INCREASE MARKET REACH AND UNLOCK NEW MONETIZATION OPPORTUNITIES

CDN INTEROPERABILITY DEFINING STANDARDS THAT CAN INCREASE MARKET REACH AND UNLOCK NEW MONETIZATION OPPORTUNITIES CDN INTEROPERABILITY DEFINING STANDARDS THAT CAN INCREASE MARKET REACH AND UNLOCK NEW MONETIZATION OPPORTUNITIES Strategic White Paper The rapid growth of over-the-top (OTT) video and multiscreen pay TV

More information

Ethernet-based Software Defined Network (SDN) Cloud Computing Research Center for Mobile Applications (CCMA), ITRI 雲 端 運 算 行 動 應 用 研 究 中 心

Ethernet-based Software Defined Network (SDN) Cloud Computing Research Center for Mobile Applications (CCMA), ITRI 雲 端 運 算 行 動 應 用 研 究 中 心 Ethernet-based Software Defined Network (SDN) Cloud Computing Research Center for Mobile Applications (CCMA), ITRI 雲 端 運 算 行 動 應 用 研 究 中 心 1 SDN Introduction Decoupling of control plane from data plane

More information

Testing Software Defined Network (SDN) For Data Center and Cloud VERYX TECHNOLOGIES

Testing Software Defined Network (SDN) For Data Center and Cloud VERYX TECHNOLOGIES Testing Software Defined Network (SDN) For Data Center and Cloud VERYX TECHNOLOGIES Table of Contents Introduction... 1 SDN - An Overview... 2 SDN: Solution Layers and its Key Requirements to be validated...

More information

Load Balancing. Final Network Exam LSNAT. Sommaire. How works a "traditional" NAT? Un article de Le wiki des TPs RSM.

Load Balancing. Final Network Exam LSNAT. Sommaire. How works a traditional NAT? Un article de Le wiki des TPs RSM. Load Balancing Un article de Le wiki des TPs RSM. PC Final Network Exam Sommaire 1 LSNAT 1.1 Deployement of LSNAT in a globally unique address space (LS-NAT) 1.2 Operation of LSNAT in conjunction with

More information

ExamPDF. Higher Quality,Better service!

ExamPDF. Higher Quality,Better service! ExamPDF Higher Quality,Better service! Q&A Exam : 1Y0-A21 Title : Basic Administration for Citrix NetScaler 9.2 Version : Demo 1 / 5 1.Scenario: An administrator is working with a Citrix consultant to

More information

Network Functions Virtualization in Home Networks

Network Functions Virtualization in Home Networks Network Functions Virtualization in Home Networks Marion Dillon Timothy Winters Abstract The current model of home networking includes relatively low- cost, failure- prone devices, requiring frequent intervention

More information

Routing & Traffic Analysis for Converged Networks. Filling the Layer 3 Gap in VoIP Management

Routing & Traffic Analysis for Converged Networks. Filling the Layer 3 Gap in VoIP Management Routing & Traffic Analysis for Converged Networks Filling the Layer 3 Gap in VoIP Management Executive Summary Voice over Internet Protocol (VoIP) is transforming corporate and consumer communications

More information

State of the Cloud DNS Report

State of the Cloud DNS Report transparency for the cloud State of the Cloud DNS Report Basic Edition August 2015 2015 Table of Contents Overview Introduction 3 Anycast vs. Unicast DNS 3 Provider Overview & Current News 4 Provider Marketshare

More information

OpenFlow: Load Balancing in enterprise networks using Floodlight Controller

OpenFlow: Load Balancing in enterprise networks using Floodlight Controller OpenFlow: Load Balancing in enterprise networks using Floodlight Controller Srinivas Govindraj, Arunkumar Jayaraman, Nitin Khanna, Kaushik Ravi Prakash srinivas.govindraj@colorado.edu, arunkumar.jayaraman@colorado.edu,

More information

Terminology. Internet Addressing System

Terminology. Internet Addressing System Terminology A local area network (LAN) is a computer network covering a small physical area, like a home, office, or small group of buildings, such as a school, or an airport. The defining characteristics

More information

TRUFFLE Broadband Bonding Network Appliance BBNA6401. A Frequently Asked Question on. Link Bonding vs. Load Balancing

TRUFFLE Broadband Bonding Network Appliance BBNA6401. A Frequently Asked Question on. Link Bonding vs. Load Balancing TRUFFLE Broadband Bonding Network Appliance BBNA6401 A Frequently Asked Question on Link Bonding vs. Load Balancing LBRvsBBNAFeb15_08b 1 Question: What's the difference between a Truffle Broadband Bonding

More information

The Evolution to Local Content Delivery

The Evolution to Local Content Delivery The Evolution to Local Content Delivery Content caching brings many benefits as operators work out how to cope with the astounding growth in demand for Internet video and other OTT content driven by Broadband

More information

Novel Systems. Extensible Networks

Novel Systems. Extensible Networks Novel Systems Active Networks Denali Extensible Networks Observations Creating/disseminating standards hard Prototyping/research Incremental deployment Computation may be cheap compared to communication

More information

Global Server Load Balancing

Global Server Load Balancing White Paper Global Server Load Balancing APV Series Application Delivery Controllers May 2011 Global Server Load Balancing Access. Security. Delivery. Introduction Scalability, high availability and performance

More information

How To Provide Qos Based Routing In The Internet

How To Provide Qos Based Routing In The Internet CHAPTER 2 QoS ROUTING AND ITS ROLE IN QOS PARADIGM 22 QoS ROUTING AND ITS ROLE IN QOS PARADIGM 2.1 INTRODUCTION As the main emphasis of the present research work is on achieving QoS in routing, hence this

More information

Why ISPs need SDN: SDN-based Network Service Chaining and Software-defined Multicast

Why ISPs need SDN: SDN-based Network Service Chaining and Software-defined Multicast Why ISPs need SDN: SDN-based Network Chaining and Software-defined Multicast ZKI Herbsttagung, Kaiserslautern, Germany, 24. Sept. 2014 Jeremias Blendin, Julius Rückert, David Hausheer Department of Electrical

More information

Scaling with Zeus Global Load Balancer

Scaling with Zeus Global Load Balancer White Paper Scaling with Zeus Global Load Balancer Zeus. Why wait Contents Introduction... 3 Server Load Balancing within a Datacenter... 3 Global Server Load Balancing between Datacenters... 3 Who might

More information

Networking Topology For Your System

Networking Topology For Your System This chapter describes the different networking topologies supported for this product, including the advantages and disadvantages of each. Select the one that best meets your needs and your network deployment.

More information

Securing Local Area Network with OpenFlow

Securing Local Area Network with OpenFlow Securing Local Area Network with OpenFlow Master s Thesis Presentation Fahad B. H. Chowdhury Supervisor: Professor Jukka Manner Advisor: Timo Kiravuo Department of Communications and Networking Aalto University

More information

Configuring Network Address Translation (NAT)

Configuring Network Address Translation (NAT) 8 Configuring Network Address Translation (NAT) Contents Overview...................................................... 8-3 Translating Between an Inside and an Outside Network........... 8-3 Local and

More information

1. Comments on reviews a. Need to avoid just summarizing web page asks you for:

1. Comments on reviews a. Need to avoid just summarizing web page asks you for: 1. Comments on reviews a. Need to avoid just summarizing web page asks you for: i. A one or two sentence summary of the paper ii. A description of the problem they were trying to solve iii. A summary of

More information

Content Delivery Networks

Content Delivery Networks Content Delivery Networks Silvano Gai Cisco Systems, USA Politecnico di Torino, IT sgai@cisco.com 1 Agenda What are Content Delivery Networks? DNS based routing Server Load Balancing Content Routers Ethical

More information

Essential Ingredients for Optimizing End User Experience Monitoring

Essential Ingredients for Optimizing End User Experience Monitoring Essential Ingredients for Optimizing End User Experience Monitoring An ENTERPRISE MANAGEMENT ASSOCIATES (EMA ) White Paper Prepared for Neustar IT MANAGEMENT RESEARCH, Table of Contents Executive Summary...1

More information

Route Control Optimize Multi-homed Connections for Performance, Load and Cost By John Bartlett January 2002

Route Control Optimize Multi-homed Connections for Performance, Load and Cost By John Bartlett January 2002 Route Control Optimize Multi-homed Connections for Performance, Load and Cost By John Bartlett January 2002 The Internet is coming of age, in large part because of its ability to open up markets and to

More information

State of the Cloud DNS Report

State of the Cloud DNS Report transparency for the cloud State of the Cloud DNS Report Basic Edition April 2015 2015 Table of Contents Overview Introduction 3 Anycast vs. Unicast DNS 3 Provider Overview & Current News 4 Provider Marketshare

More information

1.1. Abstract. 1.2. VPN Overview

1.1. Abstract. 1.2. VPN Overview 1.1. Abstract Traditionally organizations have designed their VPN networks using layer 2 WANs that provide emulated leased lines. In the last years a great variety of VPN technologies has appeared, making

More information

The Key to Cost-Effective WAN Optimization - White Paper

The Key to Cost-Effective WAN Optimization - White Paper The Key to Cost-Effective WAN Optimization - White Paper Table of Contents General...3 The Challenges of WAN Link Management in the Datacenter...3 Link Up-time and Business Continuity...3 The Active-backup

More information

A Coordinated. Enterprise Networks Software Defined. and Application Fluent Programmable Networks

A Coordinated. Enterprise Networks Software Defined. and Application Fluent Programmable Networks A Coordinated Virtual Infrastructure for SDN in Enterprise Networks Software Defined Networking (SDN), OpenFlow and Application Fluent Programmable Networks Strategic White Paper Increasing agility and

More information

Personalizing the Home Network Experience using Cloud-Based SDN

Personalizing the Home Network Experience using Cloud-Based SDN Personalizing the Home Network Experience using Cloud-Based SDN Hassan Habibi Gharakheili (UNSW) Jacob Bass (Newtworks) Luke Exton (UNSW) Vijay Sivaraman (UNSW) 1 Overview This paper is about service management

More information

Monitoring commercial cloud service providers

Monitoring commercial cloud service providers Monitoring commercial cloud service providers July Author: Lassi Kojo Supervisor: Stefan Lüders CERN openlab Summer Student Report Abstract There is a growing tendency by individuals to sign-up for public

More information

KASPERSKY DDoS PROTECTION. Protecting your business against financial and reputational losses with Kaspersky DDoS Protection

KASPERSKY DDoS PROTECTION. Protecting your business against financial and reputational losses with Kaspersky DDoS Protection KASPERSKY DDoS PROTECTION Protecting your business against financial and reputational losses A Distributed Denial of Service (DDoS) attack is one of the most popular weapons in the cybercriminals arsenal.

More information

Wedge Networks: Transparent Service Insertion in SDNs Using OpenFlow

Wedge Networks: Transparent Service Insertion in SDNs Using OpenFlow Wedge Networks: EXECUTIVE SUMMARY In this paper, we will describe a novel way to insert Wedge Network s multiple content security services (such as Anti-Virus, Anti-Spam, Web Filtering, Data Loss Prevention,

More information

DEPLOYMENT GUIDE Version 1.1. DNS Traffic Management using the BIG-IP Local Traffic Manager

DEPLOYMENT GUIDE Version 1.1. DNS Traffic Management using the BIG-IP Local Traffic Manager DEPLOYMENT GUIDE Version 1.1 DNS Traffic Management using the BIG-IP Local Traffic Manager Table of Contents Table of Contents Introducing DNS server traffic management with the BIG-IP LTM Prerequisites

More information

Chapter 6 Configuring the SSL VPN Tunnel Client and Port Forwarding

Chapter 6 Configuring the SSL VPN Tunnel Client and Port Forwarding Chapter 6 Configuring the SSL VPN Tunnel Client and Port Forwarding This chapter describes the configuration for the SSL VPN Tunnel Client and for Port Forwarding. When a remote user accesses the SSL VPN

More information

How To Make A Vpc More Secure With A Cloud Network Overlay (Network) On A Vlan) On An Openstack Vlan On A Server On A Network On A 2D (Vlan) (Vpn) On Your Vlan

How To Make A Vpc More Secure With A Cloud Network Overlay (Network) On A Vlan) On An Openstack Vlan On A Server On A Network On A 2D (Vlan) (Vpn) On Your Vlan Centec s SDN Switch Built from the Ground Up to Deliver an Optimal Virtual Private Cloud Table of Contents Virtualization Fueling New Possibilities Virtual Private Cloud Offerings... 2 Current Approaches

More information

Meeting Worldwide Demand for your Content

Meeting Worldwide Demand for your Content Meeting Worldwide Demand for your Content Evolving to a Content Delivery Network A Lucent Technologies White Paper By L. R. Beaumont 4/25/01 Meeting Worldwide Demand for your Content White Paper Table

More information

Cisco Videoscape Distribution Suite Service Broker

Cisco Videoscape Distribution Suite Service Broker Data Sheet Cisco Videoscape Distribution Suite Service Broker Product Overview Cisco Videoscape Distribution Suite Service Broker (VDS-SB) encompasses a broad range of capabilities, particularly in accelerating

More information

CS101 Lecture 19: Internetworking. What You ll Learn Today

CS101 Lecture 19: Internetworking. What You ll Learn Today CS101 Lecture 19: Internetworking Internet Protocol IP Addresses Routing Domain Name Services Aaron Stevens (azs@bu.edu) 6 March 2013 What You ll Learn Today What is the Internet? What does Internet Protocol

More information

SOFTWARE-DEFINED NETWORKING AND OPENFLOW

SOFTWARE-DEFINED NETWORKING AND OPENFLOW SOFTWARE-DEFINED NETWORKING AND OPENFLOW Eric Choi < echoi@brocade.com> Senior Manager, Service Provider Business Unit, APJ 2012 Brocade Communications Systems, Inc. EPF 7 2012/09/17 Software-Defined Networking

More information

EECS 489 Winter 2010 Midterm Exam

EECS 489 Winter 2010 Midterm Exam EECS 489 Winter 2010 Midterm Exam Name: This is an open-book, open-resources exam. Explain or show your work for each question. Your grade will be severely deducted if you don t show your work, even if

More information

Software Defined Networking & Openflow

Software Defined Networking & Openflow Software Defined Networking & Openflow Autonomic Computer Systems, HS 2015 Christopher Scherb, 01.10.2015 Overview What is Software Defined Networks? Brief summary on routing and forwarding Introduction

More information

Configuring PA Firewalls for a Layer 3 Deployment

Configuring PA Firewalls for a Layer 3 Deployment Configuring PA Firewalls for a Layer 3 Deployment Configuring PAN Firewalls for a Layer 3 Deployment Configuration Guide January 2009 Introduction The following document provides detailed step-by-step

More information

Overlay Networks. Slides adopted from Prof. Böszörményi, Distributed Systems, Summer 2004.

Overlay Networks. Slides adopted from Prof. Böszörményi, Distributed Systems, Summer 2004. Overlay Networks An overlay is a logical network on top of the physical network Routing Overlays The simplest kind of overlay Virtual Private Networks (VPN), supported by the routers If no router support

More information

GPRS / 3G Services: VPN solutions supported

GPRS / 3G Services: VPN solutions supported GPRS / 3G Services: VPN solutions supported GPRS / 3G VPN soluti An O2 White Paper An O2 White Paper Contents Page No. 3 4-6 4 5 6 6 7-10 7-8 9 9 9 10 11-14 11-12 13 13 13 14 15 16 Chapter No. 1. Executive

More information

The Value of a Content Delivery Network

The Value of a Content Delivery Network September 2010 White Paper The Value of a Content Delivery Network Table of Contents Introduction... 3 Performance... 3 The Second Generation of CDNs... 6 Conclusion... 7 About NTT America... 8 Introduction

More information

WHITEPAPER MPLS: Key Factors to Consider When Selecting Your MPLS Provider

WHITEPAPER MPLS: Key Factors to Consider When Selecting Your MPLS Provider WHITEPAPER MPLS: Key Factors to Consider When Selecting Your MPLS Provider INTRODUCTION Multiprotocol Label Switching (MPLS), once the sole domain of major corporations and telecom carriers, has gone mainstream

More information

Using TrueSpeed VNF to Test TCP Throughput in a Call Center Environment

Using TrueSpeed VNF to Test TCP Throughput in a Call Center Environment Using TrueSpeed VNF to Test TCP Throughput in a Call Center Environment TrueSpeed VNF provides network operators and enterprise users with repeatable, standards-based testing to resolve complaints about

More information

Global Load Balancing Solutions

Global Load Balancing Solutions Global Load Balancing Solutions The HTRC Group P.O. Box 2087 San Andreas, CA 95249 www.htrcgroup.com About The HTRC Group, LLC The High-tech Resource Consulting Group, LLC focuses on service provider networking,

More information

Disaster Recovery White Paper

Disaster Recovery White Paper Introduction Remote access plays a critical role in successfully executing a business recovery plan both in terms of providing access for existing remote users and accommodating the potential increase

More information

JK0-022 CompTIA Academic/E2C Security+ Certification Exam CompTIA

JK0-022 CompTIA Academic/E2C Security+ Certification Exam CompTIA JK0-022 CompTIA Academic/E2C Security+ Certification Exam CompTIA To purchase Full version of Practice exam click below; http://www.certshome.com/jk0-022-practice-test.html FOR CompTIA JK0-022 Exam Candidates

More information

Dissertation Title: SOCKS5-based Firewall Support For UDP-based Application. Author: Fung, King Pong

Dissertation Title: SOCKS5-based Firewall Support For UDP-based Application. Author: Fung, King Pong Dissertation Title: SOCKS5-based Firewall Support For UDP-based Application Author: Fung, King Pong MSc in Information Technology The Hong Kong Polytechnic University June 1999 i Abstract Abstract of dissertation

More information

Three Key Design Considerations of IP Video Surveillance Systems

Three Key Design Considerations of IP Video Surveillance Systems Three Key Design Considerations of IP Video Surveillance Systems 2012 Moxa Inc. All rights reserved. Three Key Design Considerations of IP Video Surveillance Systems Copyright Notice 2012 Moxa Inc. All

More information

MPLS: Key Factors to Consider When Selecting Your MPLS Provider Whitepaper

MPLS: Key Factors to Consider When Selecting Your MPLS Provider Whitepaper MPLS: Key Factors to Consider When Selecting Your MPLS Provider Whitepaper 2006-20011 EarthLink Business Page 1 EXECUTIVE SUMMARY Multiprotocol Label Switching (MPLS), once the sole domain of major corporations

More information