1 Agenda 14.1. Introduction, Sakari Luukkainen 21.1. Theoretical frameworks, Sakari Luukkainen 28.1. Business model design, Sakari Luukkainen 4.2. ICT in business process, Sakari Luukkainen 11.2. STOF cases, Olli Mäkinen 25.2. Cloud computing, Sakari Luukkainen 4.3. Green ICT, Jukka Nurminen 11.3. Mobile cloud computing, Sakari Luukkainen 18.3. Mobile ecosystems, Juha Winter 25.3. ICT start-up development, Aalto ACE 1.4. Summary, Sakari Luukkainen 10.4. Examination
2 Introduction term "cloud" was first used as a metaphor for the Internet, based on the cloud drawing used to depict the Internet as an abstraction of the underlying infrastructure evolution from scientific grid computing new paradigm to offer ICT services to the market - computing like electricity cloud computing providers deliver services online that are accessed from Web browser over the Internet, while the software and data are stored on servers outsourced services produced in massive centralized energy efficient automated datacenter factories cloud service market expected to grow fast (Gartner)
3 Characteristics of cloud computing
4 Characteristics of cloud computing elastical, scalable provisioning automatic management, self service provision from shared multitenant environment, better utilization rate of servers pay per use usage inpendent of terminal and location usage measurements high volume services
5 Economies of scale Cheaper MIPS (5-7 times) Need for cost efficiency of servers, globally vacancy usage cost estimated to be16 B (Kelton Research) Better utilization of computing resources (5-10% to 60-80%) Multi-tenancy: one server can serve several customers Less admin people per server (from 1:100 up to 1:10 000) Worth 1$ IT requires 8$ admin costs Source: Rolf Harms and Michael Yamartino: The Economics of the Cloud, Nov. 2010.
6 Economics of cloud centralized datacenters benefit from ecomies of scale big companies can have over million servers by cloud usage some cost sources remain, but their relative share change from fixed to variable cost (CAPEX -> OPEX) cost sources: sw and hw investment and maintenance, telecommunication, delays, updates, breaks to business process, space, electricity, cooling, security, insurance, training, support commercial offerings are generally expected to meet quality of service (QoS) requirements of customers and typically include service level agreements (SLAs) with penalty
7 Elasticity Source: Ambrust et al, Above the Clouds: A Berkeley View of Cloud Computing, Feb 2009
8 Cloud Instance Cloud instance refers to a virtual server instance from a cloud network. Cloud instance computing is highly dynamic, enabling users not to worry about how many servers can fit on a single hardware application without causing major slowdowns during peak hours. If the server grows beyond the limits of a single machine, cloud instance computing allows the cloud software to be easily expanded to span multiple machines, whether temporarily or permanently. Cloud instance computing also reduces the downtime associated with servicing hardware. A server in the cloud can be easily moved from one physical machine to another without going down. Source: http://apprenda.com/library/glossary/definition-cloud-instance-single-multi/
9 Server virtualization historically servers have been dedicated to a specific application or task and most use only small fraction of their overall processing capability virtualization is a method of running multiple independent virtual OS and applications on a single physical computer allows workload to be targeted dynamically to less used machines servers are largest power consumers of datacenters, ca. 50 % the consolidation of multiple low-load systems produce less heat and require less power and cooling market leader VMware
10 Server virtualization - shared use of resources - increased utilization rate - faster installation - reduced space - energy efficiency - faster recovery - improved maintenace Source: http://www.petri.co.il/server-virtualization-network-virtualization-storage-virtualization.htm
11 Server virtualization Case: 311 servers datacenter, 180 can be virtualized, otherwise 135 new has to be bought in 3 years, 0,4 KW / server, 0,08 / KWh Traditional CAPEX (135) 542 295 OPEX 160 650 Total 702 945 Virtualized CAPEX 220 500 OPEX 103 736 Total 324 236 ROI 217 %, savings 378 709, savings in electricity 412 070 Source: Heino 2010
12 Server virtualization To avoid vendor lock-in, the preferred way is to select an open-source cloud platform Open technologies allow the network provider to implement custom modifications to the platform Multiple options exist for open source IaaS cloud stacks: OpenNebula, Eucalyptus, CloudStack and OpenStack There are differences in licensing policy : Eucalyptus has the GNU General Public License version 3. The others have Apache License version 2. Apache License does not require releasing the modifications to the public. Other notable difference is the size, activity and governance of the community behind the project
13 Server virtualization Eucalyptus and OpenNebula have smaller communities and are controlled by a single institutes. CloudStack s community is larger, but it still largely consists of Citrix employees. OpenStack has the largest and active community, which is distributed over different institutions. Large and diverse community reduces the risk of the project being directed in a harmful direction. The positive effects are visible in, for example, the amount of supported hypervisors and networking technologies. OpenStack is widely used: in addition to smaller private clouds owners, many commercial public cloud service providers, such as Rackspace, have adopted it as their platform
14 Classification of cloud services Business model Software as a service (SaaS) e.g. Salesforce.com Platform as a Service (Paas) e.g. Google AppEngine Infrastructure as a Service (IaaS) e.g. Amazon EC2
15 Infrastructure as a Service (IaaS) evolved from virtual private server offerings customer buys computing resources from service provider as service offereded capacity capacity virtualized, scaled and automated pay per use, self service customer takes care of installing own applications, updates, load balancing and security
16 Amazon Web Services after the dot.com bubble unlike others Amazon invested to its infrastructrure first public cloud EC2 service in 2006, offers computing and storage capacity as service, as customers e.g. Pfizer, SAP, Twitter, Second Life different sized and prized instances pricing combination of fixed year fee and per hour e.g. 227 / 0,04 $ / instance, capacity can be increased by self service using credit card spot pricing for excess capacity, customer defines maximum price / instance, typically 1/3 from normal price availability 99,95 %, 4,38 hours sales / profit ca. 330 / 40 M$ in 2009, high growth
17 Platform as a Service (Paas) tools for developing, testing and maintaining applications are provided service provider use own or external IaaS environment easier and faster application development, most rutines as ready modules scalability of ready application cost efficiency, enables new entrants to application market lock-in to service provider, new competence required and security as disadvantage
18 Google AppEngine application development tools and computing resources services support only to Java and Python based applications application gets automaticly load balancing and scaling free up to 500 MB and 5 M / month page views, overdrawn capacity costs 0,1 $ / hour, transferred data 0,1-0,12 $ / GB, storage 0,15 $ / GB / month for enterprise customers 8 $ / user / month, max 1000 $ availability 99,9 % additionally SaaS GoogleApps
19 Microsoft Azure started as follower in 2010 support to.net frameworks, C++, Java, Python, Ruby and PHP based applications less limits on usage of instances capacity costs 0,12 $ / hour, transferred data 0,1-0,15 $ / GB, storage 0,15 $ / GB / month, 0,01 $ / 10 000 transactions
20 Software as a service (SaaS) customer buys plain application as service application provided through browser, no maintenance and updating required by customer - focus on business process development reliability, trustworthiness and security as disadvantage, less lock in service provider use own or external IaaS/PaaS environment customer gets own reporting and management consol by which is possible to monitor application and to add/abolish users pay per use, pay per users, flexibility in cyclical business trends larger customer base for application provider, efficient updates and deliveries focus on application development Example salesforce.com CRM applications, Gmail, YouTube, Netflix
21 Netflix - started in 1997 as a video rental business - currently largest VoD provider worldwide: 44 million customers in 41 countries - innovative company both in the sense of business and technology - new and exclusive content available to its customers - smart recommendation algorithm - support a multitude of devices - CDNs for delivery of content e.g. Akamai - uses Amazon's IaaS cloud services
22 Netflix - Netflix's own data center for registering the users and payment operations - costs: content (licensing agreements), CDN delivery, Amazon services, own data centers - revenues: from subscriptions, the more customer base, the bigger the revenue, flat rate - risks: a lot of contracts for content licensing have been signed, however, there is no guarantee that the customer base will grow (and stay) as quickly as expected, usage of public cloud platform reduce risks
23 Public, private and hybrid cloud in public cloud is services are dynamically provided over the Internet by a third-party provider like Amazon private cloud is a virtualized computing infrastructure created and managed by an organization for own internal use hybrid cloud is a cloud computing environment in which an organization produces some services inhouse and buys others from public cloud
24 Cost evaluation traditional public cloud hybrid private CAPEX 0 3 6,1 7 OPEX 77,3 22,5 28,9 31,1 Total cost 77,3 25,5 35 38,1 BCR - 15,4 6,8 5,7 Case existing traditional non virtualized datacenter, 1000 servers, cost M$ in 2 years investment and 12 years usage timeframe BCR=Benefit-to-cost ratio Source: Booz Allen Hamilton
25 Green cloud services ICT industry produce about 2% of the worldwide carbon dioxide (CO2) emissions Modern datacenters utilize waste heat and use free cooling, there are many techniques emerging to increase energy efficiency, however the incentive to apply them is currently mainly related to decreasing costs of datacenter Hosting companies and their customers could make green service level agreements, which determine relevant metrics, however there is no need for green services until the end users of web services start to demand them Every action in the Internet requires electricity: e.g. single Google search uses 0.3 watts of electricity (Glanz, 2011) - in the future some customers might be even willing to pay more for services that are proven to be green: green business models
26 Buying cloud services end up capacity discontinuity situation related to resources life cycle reasons elasticity according own market development, new entrants cost efficiency, restrictions in personnel and competences change management needs in IT administration, investing in relational technical and managerial capabilities focus on business process development and core competence trust new applications not possible traditionally
27 Utility computing The increasingly perceived vision of cloud computing as utility like electricity creates great challenges to the development of the emerging market structures The history has shown that separation of network and service has increased competition, in former monopoly, energy and telecommunications industries The markets perform in these industries more efficiently because of increased interoperability and lower switching costs The public cloud computing market is still dominated by services based on proprietary platforms and customer interfaces
28 Utility computing Under these kind of circumstances the customer expose switching costs and lock in to the cloud service provider Other observed problem, which hinders the proliferation of cloud computing, is related to trust issues between service providers and their customers. SaaS providers can easily lose their reputation, if the underlying IaaS infrastructure creates QoS or privacy problems Currently there are significant efforts to standardize customer interfaces of public cloud in order to realize interoperability and competition between various clouds The interoperability problems can also be outsourced to brokers such as RightScale and CloudSwitch
29 Future utility computing ecosystem PuC 1 PuC 2 PuC n tc 1 n sw 1 n p 1,u 1,r 1 p 2,u 2,r 2 p n,u n,r n PrC private cloud / web hotel / broker PuC public cloud PrC 1 PrC 2 PrC n c total cost u utilization rate r revenue p variable price tc telecom cost sw switching cost c 1 c 2 c n SaaS / End user
30 Future utility computing cost model Source: Raivio et al