Terena IBM Cloud Computing Pol Mac Aonghusa CTO IBM Emerging Business Incubation Center Dublin, Ireland 2009 IBM Corporation
IBM in Ireland is driving IBM leadership in Cloud Computing DUBLIN, IRELAND and ARMONK, N.Y, March 19, 2008 'Today IBM (NYSE: IBM) and the Industrial Development Agency of Ireland (IDA Ireland) announced the establishment of Europe s first Cloud Computing Center.' Center Charter Worldwide Cloud delivery infrastructure Deep skills and resources Research IBM & Partner "IBM's European hub for Cloud Computing highlights Ireland s role as an important contributor to IBM's global research, development and innovation strategy. Micheál Martin TD, Minister for Enterprise, Trade and Employment This new facility and the cloud computing model, the wealth of talent at IBM's software lab in Ireland will be accessible to not only the rest of Europe, but Africa and the Middle East as well." Steve Mills, Senior Vice President and Group Executive, IBM Software Group. 2
Dublin is part of a network of IBM Cloud Centers around the World Dublin, Ireland Seattle, WA Beijing, China Seoul, S Korea San Jose, CA US, East Coast Middle East Wuxi, China Bangalore, India Hanoi, Vietnam São Paulo, Brazil South Africa 3
Cloud Computing a computing platform for the Universe of Things Fast growth of connected devices & users Exponential growth of processing & storage demands Skyrocketing utility costs Demand for higher efficiency Standardization & Interoperability Optimised loadbalancing policies Scale rapidly and simply in a multi-tenant environment Complex collections of hardware & software A cloud is an IT delivery infrastructure with: Virtualization of hardware & software A simple interface for scheduling & provisioning resources Support to create optimised service level policies Standards for defining, managing & inter-operation of multiple clouds By 2012, 80% of Fortune 1000 enterprises will use some Cloud Computing services - Gartner 4
Cloud Computing is the culmination of a long term trend to simplify access to IT Services 1961: John McCarthy proposes computing as a utility 1961: IBM Services Bureau 1975: First inter-industry EDI standards 1981: SMTP defines the standard electronic mail service 1985: United Nations sponsors EDIFACT 1990: Berners-Lee invents the World-Wide Web 1994: CommerceNet 1998: RosettaNet 1999: i-mode mobile internet 2000: IBM BCRS 2000: UDDI 1.0; SaaS coined 2001: Dot com bubble bursts IBM Service Bureau (1961) 2005: IBM AoD 2006: Amazon EC2 2007: Google Health; force.com launch 2008: IBM ww Cloud Computing centers 1960-1970 1980 1990 2000 2010 5
Cloud Computing Logical Architecture SLA Capacity Planning Request Driven Provisioning Dynamic Scheduling Monitoring Innovation Enablement Software Development Virtual Classroom Web 2.0 Data Intensive Processing Scalable Transaction Processing Workloads Virtual Virtual Application Virtual Application Application Virtualization Virtual s Virtual Storage Virtual Networks IBM System z, p, x, BladeCenter Storage Networking Physical Hardware
Cloud - Hosting Software Development Dynamic Scheduling Monitoring China Cloud Computing Center New Enterprise Data Center built by IBM for municipal government of Wuxi, China Eleven parks across China for software development Accelerates transformation to a serviceled economy Benefits Fast deployment of Rational software development environments Up to 200K software developers, 100 companies Cost efficient shared infrastructure Request Driven Provisioning Company A Workloads Virtual Virtual Application Virtual Application Application ; Virtualization Securely isolated development and test environments Company B Company C Physical Hardware
Cloud Enabling Virtual Classrooms Dynamic Scheduling Monitoring Request Driven Provisioning Google/IBM Academic Initiative Promote open standards & Hadoop parallel computing model Jointly provide compute platform of the future Stanford Workloads Carnegie Mellon University of Washington Benefits Trains students with next generation computing skills Optimizes emerging Internet scale workloads such as search, video, audio, 3D Internet, machine learning, mobile computing Virtual Virtual Application Virtual Application Application ; Virtualization Physical Hardware
Steps towards cloud computing IBM Blue Cloud Reduce complexity and management overheads, increase efficiency Complex Scale-Out Sprawl Physical Consolidation Abstraction and Pooling SOA, Cloud Service Management, Ensembles, Windows s Unix s Switches Linux s Firewalls, Routers Management s Storage V Mainframe or Unix V Windows V Linux V Networks V Storage Virtual s, Storage, Networks Multi-System Virtualization V s V Networks We are here V Storage Ensemble Ensemble Shared network delivered services Ensemble An ensemble is a pool of like systems that is manageable as a single system It integrates compatible networked systems, virtualization, and management functions It scales from few to 1000 s of nodes, while having management complexity and cost like that of a single system essentially independent of the ensemble size 9 Page 9
Some existing ensembles Google Amazon, Yahoo, ebay, IBM: Blue Gene supercomputer System z Parallel Sysplex, up to 32 nodes SVC cluster (IBM TotalStorage San Volume Controller) XIV Nextra (IBM acquisition clustered SAN solution) Incremental Value of Storage Ensembles Autonomic management functions: (provisioning / deprovisioning, tuning, alert exception handling..) Improved staff efficiency, business adaptability for competitive advantage, position for cloud computing 10
The storage challenge where to next? 2005 2006 2007 2008 2009 2010 CAGR E-mail 269 423 673 1,119 1,512 3,553 68% Database 222 386 639 1,102 2,062 4,073 79% Unstructured 2,295 3,272 4,784 7,180 11,501 19,580 54% Total Capacity (PB) 2,786 4,081 6,096 9,401 15,076 27,206 58% Growth in stored data will continue to rise exponentially... Mankind will generate more information in the next 3 years than it did in all the previous 300,000 years How will we manage all this data? Virtualization and cloud computing with autonomic management is a key part of the solution 11
Segmentation of (Storage) Clouds Multiple dimensions of classifying offerings Product v/s Services Arms provider offerings Storage as a Service offering Public Private (Hosted or On-Premise) Standalone v/s Integrated Standalone/Direct Provides data storage, access and transfer interfaces, that are directly used by the consumer. e.g. GTS Virtual Storage Cloud Service proposal Integrated with a storage consumer E.g. Storage Ensemble integration with Ensemble for image deployment / Management E.g. Google Gmail, Docs integration of storage Functionality Data hosting, data access, data protection Content Delivery Database storage service 12
Looking Ahead Providing knowledge-enablement technologies that break the linear relationship between revenue and headcount, and exploit crowdsourcing. As a composer and integrator, using deep customer knowledge to design composite cloud solutions that combine disparate information. Service Cloud Layers 13 People Services Business Services Application Services Platform Services Infrastructure Services As a business platform provider, establishing industryspecific / standard cloud platforms through which customers can develop and offer their services. Providing leading development and management technologies for overcoming the potential complexities/ downsides of clouds. As a systems innovator, using cloud technologies to help solve out-of-space, out-of-power and lower costs. 2000 2006 2009 Static, dedicated, outsourced Network-delivered, off-premises Shared, automated, dynamic
Research Topics in Cloud Computing XTP extreme Transaction Processing Today = kilocloud ~104 105 ensemble objects 2010 > megacloud ~ 106 109 ensemble objects Advanced storage & data management Strong on parallel programming in a non-critical model Mission critical applications require robust & resilient pseudo-serial processing model New architectures to exploit massive scalability of Cloud Data intensive grid processing on Cloud Stream processing True virtualization of software for example - multi-tenancy! IPC in a mega or tera Cloud cluster! 14
Current Research Topics in Cloud Computing Device Management Explosion in Real World Aware devices producing & consuming data Nokia estimates there will be 4+ billion phones by 2010 Streaming data versus traditional data management Internet 2008 Internet 2010 exabytes (1021 bytes) zettabytes (1021 bytes) Advanced techniques required for mass back-end & front-end device management Complex ensemble management Scalable Security Services (in the Cloud) Consistent security & profiling services from device -> edge -> workload Must be dynamic & persistent across provisioning ensembles Scalability issues unless application security & profiling is virtualizable 15
Cloud Computing and Storage Storage Research Challenges Advanced caching algorithms Advanced functions for high end disk arrays Archiving Cloud computing storage ensembles Massive scaling Green Storage Solutions 16
Example R&D in Cloud: Managing Complex Ensembles Solution Composition Tool Check Out Check In Image Management Virtual image Management for prebuilt software stacks, distribution, deployment, licensing, maintenance, archival and service/ support Model based solution creation and composition tool A tool to allow administrator to quickly assemble solutions from ready-made building blocks and pre-built templates Solutions Logical Resource Topology Virtual Resources Physical Resources A Ensemble Manager Ensemble Manager Storage Ensemble Manager C1 A3 A2 C3 B2 A1 C2 B1 B3 A4 C4 Virtualization Platform Virtualization Platform B Solution Deployment and Management A1 A2 A3 A4 B1 B2 B3 C1 C2 C3 C4 Virtualization Platform Virtualization Platform Physical s Physical Ensemble Storage Ensemble C Virtualization Platform Virtualization Virtualization Platform Platform Image Manager Deploy Pre-configured Building Blocks Image Repository Image Creation Tool Isolation & connectivity management Extend isolation beyond the hypervisor boundary to networks of virtual resources Scalability and optimization integrated management and optimization of Ensembles Performance, availability & power management 17
Three Key Points Cloud: Evolution not revolution Cloud computing is a natural evolution catalysed by technological maturity and compelling business needs. IBM has proven Cloud experience - today IBM has experience implementing cloud technologies inside IBM for use carrying real workloads IBM is pursuing capabilities of tomorrow's Cloud IBM is engaged in a broad range of R&D into cloud computing technologies 18
19