Challenges in high speed packet processing

Size: px
Start display at page:

Download "Challenges in high speed packet processing"

Transcription

1 Challenges in high speed packet processing Denis Salopek University of Zagreb, Faculty of Electrical Engineering and Computing, Croatia Abstract With billions of packets traveling through the Internet and through different devices every second, the demand for fast transfer rates and even faster processing is growing. Physical layer speed is reaching its limit for 100 Gigabit Ethernet, and it is necessary to increase the processing speed of the midpoint devices (such as routers or switches). This paper brings an overview of technologies which aim to improve the flexibility and foster innovation in the domain of line-speed packet processing. Index Terms High speed networking, Packet processing, Packet forwarding, Hardware, FPGA 1. INTRODUCTION Transfer rates of 100 Gbit/s are becoming widely adopted, while vendors are under pressure from telecomm operators to introduce even faster line cards with the expectations of introducing even 400 Gbit/s in the following years [1], [2]. Commercial vendors are meeting the throughput demands by developing application specific integrated circuits (ASICs) which are employed in timing-critical datapaths. However, such an approach results in closed architectures with mostly inflexible designs. In order to gain the possibility to achieve line speeds (hypothetical maximum physical layer bitrates) in packet processing, but to also have a programmable environment for easy deployment of updates and modifications, numerous hardware and software implementations have been proposed [3] [11], of which several most important ones are covered in this paper. One of the applications for fast packet processing is route lookup which is becoming increasingly difficult and challenging with time because of the rapidly increasing number of devices connected to the internet. According to recent data 1, there are over 500 thousands different IP prefixes on the internet today. While, as a rule, transmission throughput is declared in multiples of bits per second, routers and switches have to perform roughly the same amount of processing regardless of the packet size. This makes it clear that the primary metric of measuring packet processing speed should be packets per second. As an example, while with standard 1500 byte packets a 100 Gbit/s link can transmit 8.13 millions of packets per second (Mpps), with minimum-sized (64 byte) packets the figure rises to Mpps in a single direction, or a new packet every 6.7 ns Growth of the BGP Table to Present Software defined networking (SDN) [12] is a relatively new trend in networking where all the network control is programmable and easily accessible to the user. It greatly simplifies tunneling, filtering packets, deployment of firewalls or nodes for packet inspection without the need for special hardware. It does so by inserting a layer between hardware (Network Interface Cards - NICs, queues and other network peripherals) and userspace so the traffic can be controlled directly by the user without any other obstructions in between. It represents the interface to the forwarding plane and differs from regular, non-open networking devices without such interface. Software implementations, in addition to corporations such as Cisco, are mostly being developed on general-purpose platforms (such as personal computers or servers) easily accessible to wider population, to enable every-day users the possibility to modify and upgrade router functionalities. One of the main problems with using a personal computer for packet handling and the biggest reason why doing it would be impractical and slow is packet kernel path [13] explained in more details later in this paper. The difference between using a kernel network path and avoiding it is best shown in a simple comparison in [14], where two different ways of forwarding are compared: through kernel and effectively redirecting packets from hardware directly to userspace (on FreeBSD), and without modifications to kernel, so packets follow the kernel path to the userspace application (on Linux). The maximum of around 2M packets per second for FreeBSD (around 1M packets per second for Linux) is achieved with this configuration. If the problem of kernel path is avoided, there are still some other obstacles to solve, such as data copying and interrupt handling, both of which are time consuming and need to be addressed in order to achieve high speed performance. Hardware is generally harder to upgrade so some features which are present in software implementations are not possible in hardware ones: such as the ease of updating and modifying. When the fact that the hardware development user base is much smaller is accounted, it is obvious that there is no availability to spread to a wider developer community, so these implementations are less common. The rest of this paper is organized as follows: section 2 gives an overview of current software implementations, how they work and their advantages and disadvantages. Likewise, section 3 gives the overview of current hardware implementations. Finally, section 4 summarizes the paper, gives some future work possibilities, and concludes it.

2 2. SOFTWARE IMPLEMENTATIONS Today s general-purpose computers possess high memory bandwidth, but their random memory access latency is still very high. That is why memory access is one of the biggest bottlenecks in software implementations for high speed packet processing (around 50 ns per access) and why higher number of random memory accesses per packet causes lower throughput. Because of a large number of memory accesses, kernel network path contributes a lot to this latency. Whenever a packet enters the computer via a physical interface, it must go through hardware (NIC) which stores it in the kernel memory and interrupts the CPU to notify it that a new packet is available. It then waits for its turn and gets processed by its driver when ready. After that, the driver needs to find out how to forward the packet to the upper layer (for example, there is a difference whether the packet is used for virtual LAN or for bridging). Some more time is spent on that check, and some other checks on the IP layer (error and validity checks). Depending on the packet destination (multicast, this machine or other destination), the packet is forwarded to a different socket with parameters that depend on upper layer protocol. It is obvious that some more time is spent here. Finally, after processing the packet in the transport layer (again, more time is spent), the packet is ready for the application to receive it. Then, some more validity and security checks are performed, and the packet is received. Similar path (but of course, reversed) applies for transmission, only somewhat more linear and straightforward: application sends the packet to the appropriate socket, the transport layer takes care of specific processing on the packet depending on the protocol, builds a header and sends it to kernel memory (copies it). The kernel sends the packet to the IP layer which creates its IP header and does its route lookup, after which it is sent to the link layer. Link layer takes care of scheduling and queuing packets to be sent to the NIC, and ultimately out of the computer if the destination is an external one. To avoid this complex path and high number of memory accesses, a couple of solutions already exist, such as raw sockets, the Berkeley Packet Filter (BPF) [15] and similar APIs, but they depend on certain hardware features, give unprotected access to hardware, or lack the necessary integration with the OS. Some implementations [16] suggest solutions with minimum number of high-level operations for forwarding an incoming packet: 1) direct memory access of packet from NIC to memory 2) the CPU reads packet header 3) CPU does packet processing - route lookup 4) CPU modifies the packet in memory 5) direct memory access of packet from memory to NIC 2.1 Click Click [8] is one of the first software implementations for creating and configuring network processing devices (such as routers and switches). It is implemented in the Linux kernel and that makes it possible to avoid kernel path, but it also has a user-level implementation working with BPF or any other packet socket mechanism. When Click was being developed (late 1990s), handling of multiple processes was done by using CPU interrupts. This proved to be inefficient mainly because of the livelocks they caused. To mitigate these livelocks, Click started to use polling mechanism. That way, Click polls the device for new packets and receives them when they arrive, without interrupting the processor. To send a packet, Click puts it in the transmission queue and the network device is configured to check whether there are new packets to send (as a kernel module). Click can be used to build modular routers by assembling various different elements. These elements are simple parts of a router that can perform different functionalities such as packet forwarding or packet classification individually and independently as shown in Figure 1. Figure 1. A Click example demonstrating connected elements [8] Click gives the user the possibility and simplicity to create fully functional custom routers by using predefined elements or writing their own modules without losing too much speed and functionalities. Because of that, it is often used in other software routers implementations, and one of the most

3 popular software-based routers today. In the beginnings of Click development, its maximum lossfree forwarding rate was 333K packets per second, around 4 times higher than a Linux router on the same hardware at the time. According to [17], with some modifications to include netmap s I/O libraries, Click can achieve 3.95M packets per second today. 2.2 RouteBricks RouteBricks [5] is a platform for high-speed packet processing that can use multiple interconnected general-purpose PCs, taking advantage of multiple processors and multiple cores per processor. Polling makes every CPU more efficient because no time is spent on interrupts and also it avoids packet kernel path by using kernel based Click which is directly connected to NIC. To design a router with a n ports in RouteBricks, the same number of n servers is needed. The router should receive every packet from every port and after processing them, transfer it from the input port to the required output port. Because there are multiple servers handling the packets, they preform both port switching and packet processing at the lower speed and with less CPU cost per server than any other existing software router solutions that use only one host for packet handling. Figure 2. Traditional vs. cluster-based router [16] To achieve this, all the servers are connected in clusters. This differs it from the classical router structure, as shown in Figure 2, to allow processing parallelism and communication between every server. To improve router functionality parallelism even more, it could spread processing across multiple CPUs and by using multiple memory accesses in every server. RouteBricks guarantees routing performance of 12.7 Gbit/s for 64B packets, which calculates to around 19M packets per second, and 19.4 Gbit/s minimal forwarding capabilities, i.e. around 29M packets per second for a configuration with four 8-core servers and four ports. This can scale up depending on the amount of servers and processing units, but with higher financial and power costs. 2.3 netmap netmap [9] is a framework for connecting userspace applications with hardware (NIC), for fast access to received and transmitted network packets including those used by the host stack. It is not bound to any specific hardware devices and does not need any special hardware features. To achieve high performance and speed, netmap utilises different techniques such as: metadata refactoring to lose some unnecessary packet weight fixed size packets with memory preallocation no copying of packet data, only access to the packet buffers multiple hardware queues It works by disconnecting host stack from the network adapter and assigning it to the application as shown in Figure 3. This makes packets unavailable to the host, but now they reside in the shared memory area for netmap to use. The same memory contains buffers that represent contents of the network adapter rings (NIC rings - circular arrays of buffers which represent hardware managed memory) to improve packet processing speed. Those buffers are filled depending on the status of received packets ready for processing, or for pushing the already processed packets on the transmission side. netmap claims 14.88M packets per second on a 10 Gbit/s links which is the maximum for that speed. This is the speed achieved on a single core, 900MHz computer so it is safe to assume that the faster, multi-core computers should reach even faster speeds for faster network adapters (40/100G Ethernet). Newer measurements on Intel s ixl 40G network cards indicate 32M packets per second on transmission end and 24M packets per second on receiving end with source and receiver on two different ports of the same 40G card. 2.4 DPDK DPDK [10] is, like netmap, a framework for fast userspace packet processing. It avoids kernel network path by communicating directly with the NIC and enables the userspace application to transfer packet data with minimal overhead.

4 Figure 3. netmap communication with NIC [9] Figure 4. PacketShader architecture [11] DPDK is being developed for Intel CPUs as a set of software libraries. It optimizes NIC drivers [18] by: implementing no interrupts, substituted by polling fixed size packets with memory preallocation (same as netmap) spreading objects optimally across all DRAM channels Intel claims throughput of 80M packets per second on a single 8-core processor, doubling it for the same dualprocessor on a Linux or FreeBSD userspace application. This test was performed on four 10Gbit Ethernet dual port NICs (with two NICs per processor). 2.5 PacketShader PacketShader [11] is a software router platform that uses Graphic Processing Units (GPUs) for packet processing. Graphics rendering is one of the most complicated and intensive computer operations and requires a high level of processing parallelism and computing. This, as well as GPU memory management, is what makes GPUs suitable for packet processing and enables more calculations with less strain on the CPU. The PacketShader software architecture is shown in Figure 4. PacketShader uses Linux as a host, whose network stack is not suitable for high-speed packet processing, so some modifications were being made to optimize packet I/O: Large packet buffer - for storing a greater amount of packet data which speeds up memory allocation and deallocation. Parallel hardware processing - thanks to the GPU, it is possible to process multiple packets at the same time in hardware, which greatly reduces per-packet overhead. NUMA - Non-Uniform Memory Access enables minimal movement between different types of memory. Multiple core CPU scalability - optimizations that allow linear scalability for multi-core CPU systems. PacketShader claims to achieve 40 Gbit/s packet forwarding for 64-byte packet size, reaching almost 60M packets per second. 3. HARDWARE IMPLEMENTATIONS Because of its speed and easier implementation of parallelism, working on the hardware level to gain speed is a much better choice, but with a cost of losing extensibility (as already noted above) and some other advantages of software such as memory management. Even with fast packet processing of hardware, memory access remains one of the biggest bottlenecks in hardware related development. This bottleneck is being addressed by shared memory modules and multithreaded access to those modules but more time has to be spent researching how to avoid memory accesses in packet processing. The platforms for hardware developing (such as ASICs) are mostly time consuming for their long development cycles and that is why FPGA is worth mentioning. FPGA brings fast deployment and short development cycles and that makes it suitable for hardware implementation. Intel already implemented [19] FPGA in their Xeon processors in 2014 to provide a programmable, high performance acceleration to the CPU. The FPGA is reconfigurable, so the functionalities of the FPGA can be modified depending on the task, which is intended to take over some of the load from CPU. 3.1 NetFPGA Recent trends in hardware development have shown that FPGAs are becoming more and more popular. Due to their nature of parallelism, they can be used for different tasks and more suitable for some application implementations.

5 NetFPGA [20] is an open source hardware/software platform intended for research and development of high speed hardware-based networking systems. As such, NetFPGA is not a solution for high speed packet processing, but merely a tool to accomplish such goal. The newest NetFPGA platform at this moment is NetFPGA SUME, based on Virtex-7 FPGA with the capabilities for 10, 40 and 100 Gbit/s operation and is one of the platforms that could be used for high speed processing. Being open-source, it is easily accessible and affordable platform. Figure 5 shows how the NetFPGA platform components are mutually interconnected. Some of the its features that show that it is capable of processing a high number of packets include: High-Speed Interfaces Subsystem - 30 serial links connected from the FPGA board to the Ethernet Interfaces, other extensible interfaces, interfaces for connecting the board to other boards and PCIe for connecting to the host. Memory Subsystem - 2 x 4 GB DDR3-SDRAM replaceable modules and 500 Mhz SRAM subsystems. PCIe Subsystem - 8 lane gen. PCIe 3 enabling up to 64 Gbit/s communication with the host. and other options. After processing the packet, this header is stripped. To forward received packets, SwitchBlade offers four different ways to process the received packet: Longest Prefix Matching - when a searched destination address matches multiple entries in the forwarding table, the choice for forwarding is the address with the longest subnet mask. Hashing - exact matching on the hash value based on the platform header hash value. Software processing - send the packet to the CPU to process it. User defined software processing - send the packet to the CPU for software processing but only if it matches user defined filters. SwitchBlade claims getting the maximum of 1.4M packets per second without packet dropping, and hitting that limit only because their packet generator on older, 1 Gbit/s version NetFPGA could not create packets any faster [21]. No tests were carried out on faster network cards (10G, 100G), so it is not safe to assume that the speed and performance would scale on such hardware. Figure 5. NetFPGA SUME block diagram [20] 4. CONCLUSION NetFPGA can be used in tandem with the host by using a PCIe adapter or by itself as a standalone development platform. 3.2 SwitchBlade SwitchBlade [4] is a platform for deployment of custom protocols on programmable hardware - more specifically FPGA. It is implemented on NetFPGA, but can use any type of FPGA. Besides the hardware component that is FPGA, it also uses software processing for some exceptions when the packet needs to be processed by the CPU. This actually means that SwitchBlade uses a hybrid of hardware and software approach to packet processing, but majority of this processing is done by the FPGA. Before processing, SwitchBlade first prepends a platform header to every packet. It contains information about the packet, such as its unique hash value, how it will be forwarded The Internet is expanding and the need for faster packet processing is rising, so there is always room for improvement by focusing on what is important for a machine that does the processing, and concentrating on its functionality should be a priority when doing so. In the case of packet forwarding, todays routers do not generally allow the user to modify and reconfigure packet datapath or disregard some of its unnecessary functionalities. This paper gave an overview of some of those technologies, outlining their limitations and advantages, as well as throughputs they can achieve and opportunities they open. The technologies in this field are always advancing, and speed being a priority, new ways of keeping up to date with it should be developed and researched. REFERENCES [1] C. Hermsmeyer, H. Song, R. Schlenk, R. Gemelli, and S. Bunse, Towards 100G packet processing: Challenges and technologies, Bell Labs Technical Journal, vol. 14, no. 2, pp , [2] Evolution of Ethernet speeds: what s new and what s next. apricot-2014-ethernet-speed-whats-new-and-next pdf. Created: [3] M. Bando, Y.-L. Lin, and H. J. Chao, FlashTrie: beyond 100-Gb/s IP route lookup using hash-based prefix-compressed trie, Networking, IEEE/ACM Transactions on, vol. 20, no. 4, pp , [4] M. B. Anwer, M. Motiwala, M. b. Tariq, and N. Feamster, Switchblade: a platform for rapid deployment of network protocols on programmable hardware, ACM SIGCOMM Computer Communication Review, vol. 41, no. 4, pp , 2011.

6 [5] M. Dobrescu, N. Egi, K. Argyraki, B.-G. Chun, K. Fall, G. Iannaccone, A. Knies, M. Manesh, and S. Ratnasamy, RouteBricks: exploiting parallelism to scale software routers, in Proceedings of the ACM SIGOPS 22nd symposium on Operating systems principles, pp , ACM, [6] L. Rizzo, netmap: A Novel Framework for Fast Packet I/O., in USENIX Annual Technical Conference, pp , [7] W. Sun and R. Ricci, Fast and flexible: parallel packet processing with GPUs and Click, in Proceedings of the ninth ACM/IEEE symposium on Architectures for networking and communications systems, pp , IEEE Press, [8] R. Morris, E. Kohler, J. Jannotti, and M. F. Kaashoek, The Click modular router, in ACM Transactions on Computer Systems, Citeseer, [9] L. Rizzo and M. Landi, netmap: memory mapped access to network devices, in ACM SIGCOMM Computer Communication Review, vol. 41, pp , ACM, [10] DPDK: Data Plane Development Kit. Accessed: [11] S. Han, K. Jang, K. Park, and S. Moon, PacketShader: a GPUaccelerated software router, ACM SIGCOMM Computer Communication Review, vol. 41, no. 4, pp , [12] Software-Defined Networking: The New Norm for Networks, April ONF White Paper. [13] A. K. Chimata, Path Of A Packet In The Linux Kernel Stack, [14] D. Salopek, V. Vasić, M. Zec, M. Mikuc, M. Vašarević, and V. Končar, A network testbed for commercial telecommunications product testing, in 22nd International Conference on Software, Telecommunications and Computer Networks-SoftCOM 2014, [15] S. McCanne and V. Jacobson, The BSD packet filter: A new architecture for user-level packet capture, in Proceedings of the USENIX Winter 1993 Conference Proceedings on USENIX Winter 1993 Conference Proceedings, pp. 2 2, USENIX Association, [16] K. Argyraki, S. Baset, B.-G. Chun, K. Fall, G. Iannaccone, A. Knies, E. Kohler, M. Manesh, S. Nedevschi, and S. Ratnasamy, Can software routers scale?, in Proceedings of the ACM workshop on Programmable routers for extensible services of tomorrow, pp , ACM, [17] L. Rizzo, M. Carbone, and G. Catalli, Transparent acceleration of software packet forwarding using netmap, in INFOCOM, 2012 Proceedings IEEE, pp , IEEE, [18] Intel Data Plane Development Kit (Intel DPDK) Overview Packet Processing on Intel Architecture. com/content/dam/www/public/us/en/documents/presentation/ dpdk-packet-processing-ia-overview-presentation.pdf, Accessed: [19] Disrupting the Data Center to Create the Digital Services Economy. community/itpeernetwork/datastack/blog/2014/06/18/ disrupting-the-data-center-to-create-the-digital-services-economy, [20] N. Zilberman, Y. Audzevich, A. Covington, and A. Moore, NetFPGA SUME: Toward Research Commodity 100Gb/s, [21] G. A. Covington, G. Gibb, J. W. Lockwood, and N. Mckeown, A packet generator on the NetFPGA platform, in Field Programmable Custom Computing Machines, FCCM 09., pp , IEEE, 2009.

The Lagopus SDN Software Switch. 3.1 SDN and OpenFlow. 3. Cloud Computing Technology

The Lagopus SDN Software Switch. 3.1 SDN and OpenFlow. 3. Cloud Computing Technology 3. The Lagopus SDN Software Switch Here we explain the capabilities of the new Lagopus software switch in detail, starting with the basics of SDN and OpenFlow. 3.1 SDN and OpenFlow Those engaged in network-related

More information

Intel DPDK Boosts Server Appliance Performance White Paper

Intel DPDK Boosts Server Appliance Performance White Paper Intel DPDK Boosts Server Appliance Performance Intel DPDK Boosts Server Appliance Performance Introduction As network speeds increase to 40G and above, both in the enterprise and data center, the bottlenecks

More information

Evaluating the Suitability of Server Network Cards for Software Routers

Evaluating the Suitability of Server Network Cards for Software Routers Evaluating the Suitability of Server Network Cards for Software Routers Maziar Manesh Katerina Argyraki Mihai Dobrescu Norbert Egi Kevin Fall Gianluca Iannaccone Eddie Kohler Sylvia Ratnasamy EPFL, UCLA,

More information

High-Performance IP Service Node with Layer 4 to 7 Packet Processing Features

High-Performance IP Service Node with Layer 4 to 7 Packet Processing Features UDC 621.395.31:681.3 High-Performance IP Service Node with Layer 4 to 7 Packet Processing Features VTsuneo Katsuyama VAkira Hakata VMasafumi Katoh VAkira Takeyama (Manuscript received February 27, 2001)

More information

Networking Virtualization Using FPGAs

Networking Virtualization Using FPGAs Networking Virtualization Using FPGAs Russell Tessier, Deepak Unnikrishnan, Dong Yin, and Lixin Gao Reconfigurable Computing Group Department of Electrical and Computer Engineering University of Massachusetts,

More information

Performance of Software Switching

Performance of Software Switching Performance of Software Switching Based on papers in IEEE HPSR 2011 and IFIP/ACM Performance 2011 Nuutti Varis, Jukka Manner Department of Communications and Networking (COMNET) Agenda Motivation Performance

More information

Evaluation of Switching Performance of a Virtual Software Router

Evaluation of Switching Performance of a Virtual Software Router Evaluation of Switching Performance of a Virtual Roberto Rojas-Cessa, Khondaker M. Salehin, and Komlan Egoh Abstract s are an alternative low-cost and moderate-performance router solutions implemented

More information

On Multi Gigabit Packet Capturing With Multi Core Commodity Hardware

On Multi Gigabit Packet Capturing With Multi Core Commodity Hardware On Multi Gigabit Packet Capturing With Multi Core Commodity Hardware Nicola Bonelli, Andrea Di Pietro, Stefano Giordano, and Gregorio Procissi CNIT and Università di Pisa, Pisa, Italy Abstract. Nowadays

More information

Router Architectures

Router Architectures Router Architectures An overview of router architectures. Introduction What is a Packet Switch? Basic Architectural Components Some Example Packet Switches The Evolution of IP Routers 2 1 Router Components

More information

High-Density Network Flow Monitoring

High-Density Network Flow Monitoring Petr Velan petr.velan@cesnet.cz High-Density Network Flow Monitoring IM2015 12 May 2015, Ottawa Motivation What is high-density flow monitoring? Monitor high traffic in as little rack units as possible

More information

Windows Server Performance Monitoring

Windows Server Performance Monitoring Spot server problems before they are noticed The system s really slow today! How often have you heard that? Finding the solution isn t so easy. The obvious questions to ask are why is it running slowly

More information

HANIC 100G: Hardware accelerator for 100 Gbps network traffic monitoring

HANIC 100G: Hardware accelerator for 100 Gbps network traffic monitoring CESNET Technical Report 2/2014 HANIC 100G: Hardware accelerator for 100 Gbps network traffic monitoring VIKTOR PUš, LUKÁš KEKELY, MARTIN ŠPINLER, VÁCLAV HUMMEL, JAN PALIČKA Received 3. 10. 2014 Abstract

More information

Parallel Firewalls on General-Purpose Graphics Processing Units

Parallel Firewalls on General-Purpose Graphics Processing Units Parallel Firewalls on General-Purpose Graphics Processing Units Manoj Singh Gaur and Vijay Laxmi Kamal Chandra Reddy, Ankit Tharwani, Ch.Vamshi Krishna, Lakshminarayanan.V Department of Computer Engineering

More information

ncap: Wire-speed Packet Capture and Transmission

ncap: Wire-speed Packet Capture and Transmission ncap: Wire-speed Packet Capture and Transmission L. Deri ntop.org Pisa Italy deri@ntop.org Abstract With the increasing network speed, it is no longer possible to capture and transmit network packets at

More information

ECE 578 Term Paper Network Security through IP packet Filtering

ECE 578 Term Paper Network Security through IP packet Filtering ECE 578 Term Paper Network Security through IP packet Filtering Cheedu Venugopal Reddy Dept of Electrical Eng and Comp science Oregon State University Bin Cao Dept of electrical Eng and Comp science Oregon

More information

Cray Gemini Interconnect. Technical University of Munich Parallel Programming Class of SS14 Denys Sobchyshak

Cray Gemini Interconnect. Technical University of Munich Parallel Programming Class of SS14 Denys Sobchyshak Cray Gemini Interconnect Technical University of Munich Parallel Programming Class of SS14 Denys Sobchyshak Outline 1. Introduction 2. Overview 3. Architecture 4. Gemini Blocks 5. FMA & BTA 6. Fault tolerance

More information

Next Generation Operating Systems

Next Generation Operating Systems Next Generation Operating Systems Zeljko Susnjar, Cisco CTG June 2015 The end of CPU scaling Future computing challenges Power efficiency Performance == parallelism Cisco Confidential 2 Paradox of the

More information

Open-source routing at 10Gb/s

Open-source routing at 10Gb/s Open-source routing at Gb/s Olof Hagsand, Robert Olsson and Bengt Gördén, Royal Institute of Technology (KTH), Sweden Email: {olofh, gorden}@kth.se Uppsala University, Uppsala, Sweden Email: robert.olsson@its.uu.se

More information

High-Density Network Flow Monitoring

High-Density Network Flow Monitoring High-Density Network Flow Monitoring Petr Velan CESNET, z.s.p.o. Zikova 4, 160 00 Praha 6, Czech Republic petr.velan@cesnet.cz Viktor Puš CESNET, z.s.p.o. Zikova 4, 160 00 Praha 6, Czech Republic pus@cesnet.cz

More information

White Paper. Innovate Telecom Services with NFV and SDN

White Paper. Innovate Telecom Services with NFV and SDN White Paper Innovate Telecom Services with NFV and SDN 2 NEXCOM White Paper As telecommunications companies seek to expand beyond telecommunications services to data services, they find their purposebuilt

More information

Wire-speed Packet Capture and Transmission

Wire-speed Packet Capture and Transmission Wire-speed Packet Capture and Transmission Luca Deri Packet Capture: Open Issues Monitoring low speed (100 Mbit) networks is already possible using commodity hardware and tools based on libpcap.

More information

Welcome to the Dawn of Open-Source Networking. Linux IP Routers Bob Gilligan gilligan@vyatta.com

Welcome to the Dawn of Open-Source Networking. Linux IP Routers Bob Gilligan gilligan@vyatta.com Welcome to the Dawn of Open-Source Networking. Linux IP Routers Bob Gilligan gilligan@vyatta.com Outline About Vyatta: Open source project, and software product Areas we re working on or interested in

More information

D1.2 Network Load Balancing

D1.2 Network Load Balancing D1. Network Load Balancing Ronald van der Pol, Freek Dijkstra, Igor Idziejczak, and Mark Meijerink SARA Computing and Networking Services, Science Park 11, 9 XG Amsterdam, The Netherlands June ronald.vanderpol@sara.nl,freek.dijkstra@sara.nl,

More information

Xeon+FPGA Platform for the Data Center

Xeon+FPGA Platform for the Data Center Xeon+FPGA Platform for the Data Center ISCA/CARL 2015 PK Gupta, Director of Cloud Platform Technology, DCG/CPG Overview Data Center and Workloads Xeon+FPGA Accelerator Platform Applications and Eco-system

More information

Content-Aware Load Balancing using Direct Routing for VOD Streaming Service

Content-Aware Load Balancing using Direct Routing for VOD Streaming Service Content-Aware Load Balancing using Direct Routing for VOD Streaming Service Young-Hwan Woo, Jin-Wook Chung, Seok-soo Kim Dept. of Computer & Information System, Geo-chang Provincial College, Korea School

More information

Stream Processing on GPUs Using Distributed Multimedia Middleware

Stream Processing on GPUs Using Distributed Multimedia Middleware Stream Processing on GPUs Using Distributed Multimedia Middleware Michael Repplinger 1,2, and Philipp Slusallek 1,2 1 Computer Graphics Lab, Saarland University, Saarbrücken, Germany 2 German Research

More information

Packet-based Network Traffic Monitoring and Analysis with GPUs

Packet-based Network Traffic Monitoring and Analysis with GPUs Packet-based Network Traffic Monitoring and Analysis with GPUs Wenji Wu, Phil DeMar wenji@fnal.gov, demar@fnal.gov GPU Technology Conference 2014 March 24-27, 2014 SAN JOSE, CALIFORNIA Background Main

More information

基 於 SDN 與 可 程 式 化 硬 體 架 構 之 雲 端 網 路 系 統 交 換 器

基 於 SDN 與 可 程 式 化 硬 體 架 構 之 雲 端 網 路 系 統 交 換 器 基 於 SDN 與 可 程 式 化 硬 體 架 構 之 雲 端 網 路 系 統 交 換 器 楊 竹 星 教 授 國 立 成 功 大 學 電 機 工 程 學 系 Outline Introduction OpenFlow NetFPGA OpenFlow Switch on NetFPGA Development Cases Conclusion 2 Introduction With the proposal

More information

Quantifying the Performance Degradation of IPv6 for TCP in Windows and Linux Networking

Quantifying the Performance Degradation of IPv6 for TCP in Windows and Linux Networking Quantifying the Performance Degradation of IPv6 for TCP in Windows and Linux Networking Burjiz Soorty School of Computing and Mathematical Sciences Auckland University of Technology Auckland, New Zealand

More information

Monitoring high-speed networks using ntop. Luca Deri <deri@ntop.org>

Monitoring high-speed networks using ntop. Luca Deri <deri@ntop.org> Monitoring high-speed networks using ntop Luca Deri 1 Project History Started in 1997 as monitoring application for the Univ. of Pisa 1998: First public release v 0.4 (GPL2) 1999-2002:

More information

ADVANCED NETWORK CONFIGURATION GUIDE

ADVANCED NETWORK CONFIGURATION GUIDE White Paper ADVANCED NETWORK CONFIGURATION GUIDE CONTENTS Introduction 1 Terminology 1 VLAN configuration 2 NIC Bonding configuration 3 Jumbo frame configuration 4 Other I/O high availability options 4

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

Oracle SDN Performance Acceleration with Software-Defined Networking

Oracle SDN Performance Acceleration with Software-Defined Networking Oracle SDN Performance Acceleration with Software-Defined Networking Oracle SDN, which delivers software-defined networking, boosts application performance and management flexibility by dynamically connecting

More information

Cisco Integrated Services Routers Performance Overview

Cisco Integrated Services Routers Performance Overview Integrated Services Routers Performance Overview What You Will Learn The Integrated Services Routers Generation 2 (ISR G2) provide a robust platform for delivering WAN services, unified communications,

More information

Enhance Service Delivery and Accelerate Financial Applications with Consolidated Market Data

Enhance Service Delivery and Accelerate Financial Applications with Consolidated Market Data White Paper Enhance Service Delivery and Accelerate Financial Applications with Consolidated Market Data What You Will Learn Financial market technology is advancing at a rapid pace. The integration of

More information

High-performance vswitch of the user, by the user, for the user

High-performance vswitch of the user, by the user, for the user A bird in cloud High-performance vswitch of the user, by the user, for the user Yoshihiro Nakajima, Wataru Ishida, Tomonori Fujita, Takahashi Hirokazu, Tomoya Hibi, Hitoshi Matsutahi, Katsuhiro Shimano

More information

MoonGen. A Scriptable High-Speed Packet Generator Design and Implementation. Paul Emmerich. January 30th, 2016 FOSDEM 2016

MoonGen. A Scriptable High-Speed Packet Generator Design and Implementation. Paul Emmerich. January 30th, 2016 FOSDEM 2016 MoonGen A Scriptable High-Speed Packet Generator Design and Implementation Paul Emmerich January 30th, 2016 FOSDEM 2016 Chair for Network Architectures and Services Department of Informatics Paul Emmerich

More information

Architecture of distributed network processors: specifics of application in information security systems

Architecture of distributed network processors: specifics of application in information security systems Architecture of distributed network processors: specifics of application in information security systems V.Zaborovsky, Politechnical University, Sait-Petersburg, Russia vlad@neva.ru 1. Introduction Modern

More information

How To Improve Performance On A Linux Based Router

How To Improve Performance On A Linux Based Router Linux Based Router Over 10GE LAN Cheng Cui, Chui-hui Chiu, and Lin Xue Department of Computer Science Louisiana State University, LA USA Abstract High speed routing with 10Gbps link speed is still very

More information

Big Data Technologies for Ultra-High-Speed Data Transfer and Processing

Big Data Technologies for Ultra-High-Speed Data Transfer and Processing White Paper Intel Xeon Processor E5 Family Big Data Analytics Cloud Computing Solutions Big Data Technologies for Ultra-High-Speed Data Transfer and Processing Using Technologies from Aspera and Intel

More information

OpenFlow with Intel 82599. Voravit Tanyingyong, Markus Hidell, Peter Sjödin

OpenFlow with Intel 82599. Voravit Tanyingyong, Markus Hidell, Peter Sjödin OpenFlow with Intel 82599 Voravit Tanyingyong, Markus Hidell, Peter Sjödin Outline Background Goal Design Experiment and Evaluation Conclusion OpenFlow SW HW Open up commercial network hardware for experiment

More information

Optimizing Data Center Networks for Cloud Computing

Optimizing Data Center Networks for Cloud Computing PRAMAK 1 Optimizing Data Center Networks for Cloud Computing Data Center networks have evolved over time as the nature of computing changed. They evolved to handle the computing models based on main-frames,

More information

50. DFN Betriebstagung

50. DFN Betriebstagung 50. DFN Betriebstagung IPS Serial Clustering in 10GbE Environment Tuukka Helander, Stonesoft Germany GmbH Frank Brüggemann, RWTH Aachen Slide 1 Agenda Introduction Stonesoft clustering Firewall parallel

More information

Local-Area Network -LAN

Local-Area Network -LAN Computer Networks A group of two or more computer systems linked together. There are many [types] of computer networks: Peer To Peer (workgroups) The computers are connected by a network, however, there

More information

The virtualization of SAP environments to accommodate standardization and easier management is gaining momentum in data centers.

The virtualization of SAP environments to accommodate standardization and easier management is gaining momentum in data centers. White Paper Virtualized SAP: Optimize Performance with Cisco Data Center Virtual Machine Fabric Extender and Red Hat Enterprise Linux and Kernel-Based Virtual Machine What You Will Learn The virtualization

More information

- An Essential Building Block for Stable and Reliable Compute Clusters

- An Essential Building Block for Stable and Reliable Compute Clusters Ferdinand Geier ParTec Cluster Competence Center GmbH, V. 1.4, March 2005 Cluster Middleware - An Essential Building Block for Stable and Reliable Compute Clusters Contents: Compute Clusters a Real Alternative

More information

VMWARE WHITE PAPER 1

VMWARE WHITE PAPER 1 1 VMWARE WHITE PAPER Introduction This paper outlines the considerations that affect network throughput. The paper examines the applications deployed on top of a virtual infrastructure and discusses the

More information

Performance Evaluation of Linux Bridge

Performance Evaluation of Linux Bridge Performance Evaluation of Linux Bridge James T. Yu School of Computer Science, Telecommunications, and Information System (CTI) DePaul University ABSTRACT This paper studies a unique network feature, Ethernet

More information

Open, Elastic Provisioning of Hardware Acceleration in NFV Environments

Open, Elastic Provisioning of Hardware Acceleration in NFV Environments Open, Elastic Provisioning of Hardware Acceleration in NFV Environments Leonhard Nobach, M. Sc. lnobach@ps.tu-darmstadt.de Prof. Dr. David Hausheer hausheer@ps.tu-darmstadt.de Netsys 2015 SDNFlex Workshop

More information

Sockets vs. RDMA Interface over 10-Gigabit Networks: An In-depth Analysis of the Memory Traffic Bottleneck

Sockets vs. RDMA Interface over 10-Gigabit Networks: An In-depth Analysis of the Memory Traffic Bottleneck Sockets vs. RDMA Interface over 1-Gigabit Networks: An In-depth Analysis of the Memory Traffic Bottleneck Pavan Balaji Hemal V. Shah D. K. Panda Network Based Computing Lab Computer Science and Engineering

More information

A Low Latency Library in FPGA Hardware for High Frequency Trading (HFT)

A Low Latency Library in FPGA Hardware for High Frequency Trading (HFT) A Low Latency Library in FPGA Hardware for High Frequency Trading (HFT) John W. Lockwood, Adwait Gupte, Nishit Mehta (Algo-Logic Systems) Michaela Blott, Tom English, Kees Vissers (Xilinx) August 22, 2012,

More information

An Oracle Technical White Paper November 2011. Oracle Solaris 11 Network Virtualization and Network Resource Management

An Oracle Technical White Paper November 2011. Oracle Solaris 11 Network Virtualization and Network Resource Management An Oracle Technical White Paper November 2011 Oracle Solaris 11 Network Virtualization and Network Resource Management Executive Overview... 2 Introduction... 2 Network Virtualization... 2 Network Resource

More information

Network Traffic Monitoring and Analysis with GPUs

Network Traffic Monitoring and Analysis with GPUs Network Traffic Monitoring and Analysis with GPUs Wenji Wu, Phil DeMar wenji@fnal.gov, demar@fnal.gov GPU Technology Conference 2013 March 18-21, 2013 SAN JOSE, CALIFORNIA Background Main uses for network

More information

GPU-Based Network Traffic Monitoring & Analysis Tools

GPU-Based Network Traffic Monitoring & Analysis Tools GPU-Based Network Traffic Monitoring & Analysis Tools Wenji Wu; Phil DeMar wenji@fnal.gov, demar@fnal.gov CHEP 2013 October 17, 2013 Coarse Detailed Background Main uses for network traffic monitoring

More information

Computer Network. Interconnected collection of autonomous computers that are able to exchange information

Computer Network. Interconnected collection of autonomous computers that are able to exchange information Introduction Computer Network. Interconnected collection of autonomous computers that are able to exchange information No master/slave relationship between the computers in the network Data Communications.

More information

Data Center and Cloud Computing Market Landscape and Challenges

Data Center and Cloud Computing Market Landscape and Challenges Data Center and Cloud Computing Market Landscape and Challenges Manoj Roge, Director Wired & Data Center Solutions Xilinx Inc. #OpenPOWERSummit 1 Outline Data Center Trends Technology Challenges Solution

More information

Saisei and Intel Maximizing WAN Bandwidth

Saisei and Intel Maximizing WAN Bandwidth Intel Network Builders Saisei Solution Brief Intel Xeon Processors Saisei and Intel Maximizing WAN Bandwidth Introduction Despite the increased capacity available on WAN links1, service providers and enterprises

More information

Programmable Networking with Open vswitch

Programmable Networking with Open vswitch Programmable Networking with Open vswitch Jesse Gross LinuxCon September, 2013 2009 VMware Inc. All rights reserved Background: The Evolution of Data Centers Virtualization has created data center workloads

More information

The proliferation of the raw processing

The proliferation of the raw processing TECHNOLOGY CONNECTED Advances with System Area Network Speeds Data Transfer between Servers with A new network switch technology is targeted to answer the phenomenal demands on intercommunication transfer

More information

Reconfigurable Architecture Requirements for Co-Designed Virtual Machines

Reconfigurable Architecture Requirements for Co-Designed Virtual Machines Reconfigurable Architecture Requirements for Co-Designed Virtual Machines Kenneth B. Kent University of New Brunswick Faculty of Computer Science Fredericton, New Brunswick, Canada ken@unb.ca Micaela Serra

More information

Datacenter Operating Systems

Datacenter Operating Systems Datacenter Operating Systems CSE451 Simon Peter With thanks to Timothy Roscoe (ETH Zurich) Autumn 2015 This Lecture What s a datacenter Why datacenters Types of datacenters Hyperscale datacenters Major

More information

Building an Accelerated and Energy-Efficient Traffic Monitor onto the NetFPGA platform

Building an Accelerated and Energy-Efficient Traffic Monitor onto the NetFPGA platform Building an Accelerated and Energy-Efficient Traffic Monitor onto the NetFPGA platform Alfio Lombardo Department of Computer Science and Telecommunication Engineering V.le A. Doria, 6 95125, Catania, Italy

More information

Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging

Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging Achieving Nanosecond Latency Between Applications with IPC Shared Memory Messaging In some markets and scenarios where competitive advantage is all about speed, speed is measured in micro- and even nano-seconds.

More information

The Design and Implementation of Content Switch On IXP12EB

The Design and Implementation of Content Switch On IXP12EB The Design and Implementation of Content Switch On IXP12EB Thesis Proposal by Longhua Li Computer Science Department University of Colorado at Colorado Springs 5/15/2001 Approved by: Dr. Edward Chow (Advisor)

More information

Software Defined Networking

Software Defined Networking Software Defined Networking Richard T. B. Ma School of Computing National University of Singapore Material from: Scott Shenker (UC Berkeley), Nick McKeown (Stanford), Jennifer Rexford (Princeton) CS 4226:

More information

Gigabit Ethernet Packet Capture. User s Guide

Gigabit Ethernet Packet Capture. User s Guide Gigabit Ethernet Packet Capture User s Guide Copyrights Copyright 2008 CACE Technologies, Inc. All rights reserved. This document may not, in whole or part, be: copied; photocopied; reproduced; translated;

More information

From Ethernet Ubiquity to Ethernet Convergence: The Emergence of the Converged Network Interface Controller

From Ethernet Ubiquity to Ethernet Convergence: The Emergence of the Converged Network Interface Controller White Paper From Ethernet Ubiquity to Ethernet Convergence: The Emergence of the Converged Network Interface Controller The focus of this paper is on the emergence of the converged network interface controller

More information

How To Monitor And Test An Ethernet Network On A Computer Or Network Card

How To Monitor And Test An Ethernet Network On A Computer Or Network Card 3. MONITORING AND TESTING THE ETHERNET NETWORK 3.1 Introduction The following parameters are covered by the Ethernet performance metrics: Latency (delay) the amount of time required for a frame to travel

More information

Design Issues in a Bare PC Web Server

Design Issues in a Bare PC Web Server Design Issues in a Bare PC Web Server Long He, Ramesh K. Karne, Alexander L. Wijesinha, Sandeep Girumala, and Gholam H. Khaksari Department of Computer & Information Sciences, Towson University, 78 York

More information

SDN software switch Lagopus and NFV enabled software node

SDN software switch Lagopus and NFV enabled software node SDN software switch Lagopus and NFV enabled software node Kazuaki OBANA NTT Network Innovation Laboratories SDN software switch Lagopus 1 Motivation Agile and flexible networking Full automation in provisioning,

More information

Intel Data Direct I/O Technology (Intel DDIO): A Primer >

Intel Data Direct I/O Technology (Intel DDIO): A Primer > Intel Data Direct I/O Technology (Intel DDIO): A Primer > Technical Brief February 2012 Revision 1.0 Legal Statements INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE,

More information

How Router Technology Shapes Inter-Cloud Computing Service Architecture for The Future Internet

How Router Technology Shapes Inter-Cloud Computing Service Architecture for The Future Internet How Router Technology Shapes Inter-Cloud Computing Service Architecture for The Future Internet Professor Jiann-Liang Chen Friday, September 23, 2011 Wireless Networks and Evolutional Communications Laboratory

More information

Optimizing GPU-based application performance for the HP for the HP ProLiant SL390s G7 server

Optimizing GPU-based application performance for the HP for the HP ProLiant SL390s G7 server Optimizing GPU-based application performance for the HP for the HP ProLiant SL390s G7 server Technology brief Introduction... 2 GPU-based computing... 2 ProLiant SL390s GPU-enabled architecture... 2 Optimizing

More information

I/O Virtualization Using Mellanox InfiniBand And Channel I/O Virtualization (CIOV) Technology

I/O Virtualization Using Mellanox InfiniBand And Channel I/O Virtualization (CIOV) Technology I/O Virtualization Using Mellanox InfiniBand And Channel I/O Virtualization (CIOV) Technology Reduce I/O cost and power by 40 50% Reduce I/O real estate needs in blade servers through consolidation Maintain

More information

COMPUTING. Centellis Virtualization Platform An open hardware and software platform for implementing virtualized applications

COMPUTING. Centellis Virtualization Platform An open hardware and software platform for implementing virtualized applications COMPUTING Data Sheet Centellis VP provides the hardware and software platform to deploy carrier grade virtualized applications. Application virtualization software framework based on industry standard

More information

How Solace Message Routers Reduce the Cost of IT Infrastructure

How Solace Message Routers Reduce the Cost of IT Infrastructure How Message Routers Reduce the Cost of IT Infrastructure This paper explains how s innovative solution can significantly reduce the total cost of ownership of your messaging middleware platform and IT

More information

Cloud Storage. Parallels. Performance Benchmark Results. White Paper. www.parallels.com

Cloud Storage. Parallels. Performance Benchmark Results. White Paper. www.parallels.com Parallels Cloud Storage White Paper Performance Benchmark Results www.parallels.com Table of Contents Executive Summary... 3 Architecture Overview... 3 Key Features... 4 No Special Hardware Requirements...

More information

PCI Express* Ethernet Networking

PCI Express* Ethernet Networking White Paper Intel PRO Network Adapters Network Performance Network Connectivity Express* Ethernet Networking Express*, a new third-generation input/output (I/O) standard, allows enhanced Ethernet network

More information

Lustre Networking BY PETER J. BRAAM

Lustre Networking BY PETER J. BRAAM Lustre Networking BY PETER J. BRAAM A WHITE PAPER FROM CLUSTER FILE SYSTEMS, INC. APRIL 2007 Audience Architects of HPC clusters Abstract This paper provides architects of HPC clusters with information

More information

UPPER LAYER SWITCHING

UPPER LAYER SWITCHING 52-20-40 DATA COMMUNICATIONS MANAGEMENT UPPER LAYER SWITCHING Gilbert Held INSIDE Upper Layer Operations; Address Translation; Layer 3 Switching; Layer 4 Switching OVERVIEW The first series of LAN switches

More information

Benchmarking Virtual Switches in OPNFV draft-vsperf-bmwg-vswitch-opnfv-00. Maryam Tahhan Al Morton

Benchmarking Virtual Switches in OPNFV draft-vsperf-bmwg-vswitch-opnfv-00. Maryam Tahhan Al Morton Benchmarking Virtual Switches in OPNFV draft-vsperf-bmwg-vswitch-opnfv-00 Maryam Tahhan Al Morton Introduction Maryam Tahhan Network Software Engineer Intel Corporation (Shannon Ireland). VSPERF project

More information

TCP Servers: Offloading TCP Processing in Internet Servers. Design, Implementation, and Performance

TCP Servers: Offloading TCP Processing in Internet Servers. Design, Implementation, and Performance TCP Servers: Offloading TCP Processing in Internet Servers. Design, Implementation, and Performance M. Rangarajan, A. Bohra, K. Banerjee, E.V. Carrera, R. Bianchini, L. Iftode, W. Zwaenepoel. Presented

More information

Flow processing and the rise of the middle.

Flow processing and the rise of the middle. Flow processing and the rise of the middle. Mark Handley, UCL With acknowledgments to Michio Honda, Laurent Mathy, Costin Raiciu, Olivier Bonaventure, and Felipe Huici. Part 1 Today s Internet Protocol

More information

Data Center Networks and Basic Switching Technologies

Data Center Networks and Basic Switching Technologies Data Center Networks and Basic Switching Technologies Hakim Weatherspoon Assistant Professor, Dept of Computer Science CS 5413: High Performance Systems and Networking September 15, 2014 Slides used and

More information

Brocade Solution for EMC VSPEX Server Virtualization

Brocade Solution for EMC VSPEX Server Virtualization Reference Architecture Brocade Solution Blueprint Brocade Solution for EMC VSPEX Server Virtualization Microsoft Hyper-V for 50 & 100 Virtual Machines Enabled by Microsoft Hyper-V, Brocade ICX series switch,

More information

Influence of Load Balancing on Quality of Real Time Data Transmission*

Influence of Load Balancing on Quality of Real Time Data Transmission* SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol. 6, No. 3, December 2009, 515-524 UDK: 004.738.2 Influence of Load Balancing on Quality of Real Time Data Transmission* Nataša Maksić 1,a, Petar Knežević 2,

More information

Building a Highly Available and Scalable Web Farm

Building a Highly Available and Scalable Web Farm Page 1 of 10 MSDN Home > MSDN Library > Deployment Rate this page: 10 users 4.9 out of 5 Building a Highly Available and Scalable Web Farm Duwamish Online Paul Johns and Aaron Ching Microsoft Developer

More information

Wireshark in a Multi-Core Environment Using Hardware Acceleration Presenter: Pete Sanders, Napatech Inc. Sharkfest 2009 Stanford University

Wireshark in a Multi-Core Environment Using Hardware Acceleration Presenter: Pete Sanders, Napatech Inc. Sharkfest 2009 Stanford University Wireshark in a Multi-Core Environment Using Hardware Acceleration Presenter: Pete Sanders, Napatech Inc. Sharkfest 2009 Stanford University Napatech - Sharkfest 2009 1 Presentation Overview About Napatech

More information

Enabling Technologies for Distributed Computing

Enabling Technologies for Distributed Computing Enabling Technologies for Distributed Computing Dr. Sanjay P. Ahuja, Ph.D. Fidelity National Financial Distinguished Professor of CIS School of Computing, UNF Multi-core CPUs and Multithreading Technologies

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

SDN and NFV Open Source Initiatives. Systematic SDN and NFV Workshop Challenges, Opportunities and Potential Impact

SDN and NFV Open Source Initiatives. Systematic SDN and NFV Workshop Challenges, Opportunities and Potential Impact SDN and NFV Open Source Initiatives Systematic SDN and NFV Workshop Challenges, Opportunities and Potential Impact May 19, 2014 Eric CARMES 6WIND Founder and CEO SPEED MATTERS V1.0. All rights reserved.

More information

PCI Express High Speed Networks. Complete Solution for High Speed Networking

PCI Express High Speed Networks. Complete Solution for High Speed Networking PCI Express High Speed Networks Complete Solution for High Speed Networking Ultra Low Latency Ultra High Throughput Maximizing application performance is a combination of processing, communication, and

More information

Enabling Practical SDN Security Applications with OFX (The OpenFlow extension Framework)

Enabling Practical SDN Security Applications with OFX (The OpenFlow extension Framework) Enabling Practical SDN Security Applications with OFX (The OpenFlow extension Framework) John Sonchack, Adam J. Aviv, Eric Keller, and Jonathan M. Smith Outline Introduction Overview of OFX Using OFX Benchmarks

More information

Improved Forwarding Architecture and Resource Management for Multi-Core Software Routers

Improved Forwarding Architecture and Resource Management for Multi-Core Software Routers Improved Forwarding Architecture and Resource Management for Multi-Core Software Routers Norbert Egi, Adam Greenhalgh, Mark Handley, Gianluca Iannaccone, Maziar Manesh, Laurent Mathy, Sylvia Ratnasamy

More information

Boosting Data Transfer with TCP Offload Engine Technology

Boosting Data Transfer with TCP Offload Engine Technology Boosting Data Transfer with TCP Offload Engine Technology on Ninth-Generation Dell PowerEdge Servers TCP/IP Offload Engine () technology makes its debut in the ninth generation of Dell PowerEdge servers,

More information

Quantum StorNext. Product Brief: Distributed LAN Client

Quantum StorNext. Product Brief: Distributed LAN Client Quantum StorNext Product Brief: Distributed LAN Client NOTICE This product brief may contain proprietary information protected by copyright. Information in this product brief is subject to change without

More information

Overlapping Data Transfer With Application Execution on Clusters

Overlapping Data Transfer With Application Execution on Clusters Overlapping Data Transfer With Application Execution on Clusters Karen L. Reid and Michael Stumm reid@cs.toronto.edu stumm@eecg.toronto.edu Department of Computer Science Department of Electrical and Computer

More information

Introduction to Intel Ethernet Flow Director and Memcached Performance

Introduction to Intel Ethernet Flow Director and Memcached Performance White Paper Intel Ethernet Flow Director Introduction to Intel Ethernet Flow Director and Memcached Performance Problem Statement Bob Metcalfe, then at Xerox PARC, invented Ethernet in 1973, over forty

More information

Removing Performance Bottlenecks in Databases with Red Hat Enterprise Linux and Violin Memory Flash Storage Arrays. Red Hat Performance Engineering

Removing Performance Bottlenecks in Databases with Red Hat Enterprise Linux and Violin Memory Flash Storage Arrays. Red Hat Performance Engineering Removing Performance Bottlenecks in Databases with Red Hat Enterprise Linux and Violin Memory Flash Storage Arrays Red Hat Performance Engineering Version 1.0 August 2013 1801 Varsity Drive Raleigh NC

More information

TCP Offload Engines. As network interconnect speeds advance to Gigabit. Introduction to

TCP Offload Engines. As network interconnect speeds advance to Gigabit. Introduction to Introduction to TCP Offload Engines By implementing a TCP Offload Engine (TOE) in high-speed computing environments, administrators can help relieve network bottlenecks and improve application performance.

More information