- Количество слайдов: 39
Cloud Computing Trends and Directions Prepared for IBM – June 13, 2008 David Cearley Tom Bittman Yefim Natis Notes accompany this presentation. Please select Notes Page view. These materials can be reproduced only with written approval from Gartner. Such approvals must be requested via e-mail: vendor. [email protected] com. Gartner is a registered trademark of Gartner, Inc. or its affiliates.
What's Feeding the Cloud Computing Phenomenon? Alternate Client Devices: Explosion of form factors, mobility, connectedness Infrastructure Technologies: Virtualization and automation Industrialization of IT: Standardization and commoditization (e. g. , e -mail), open source Cloud Web Internet Elastic Services Data Center Pressures: Growing costs of power and space, server sprawl Application Technologies: From parallel processing (grid, Map. Reduce, Hadoop) to Web 2. 0, SOA Information Web Platforms and E-Commerce Applications: Connectivity Elastically scalable global class 1980 1990 2000 2010 2020 infrastructure and Networking: mashable services Data-Intensive Growth in built on WOA (e. g. , Business Model: Applications: Advertising REST, RSS/Atom) From massively parallel connectivity and subsidized, (e. g. , Google) to large bandwidth through venture funding the Internet data files (e. g. , You. Tube) for service model
What Is Cloud Computing? Myths to Explore Cloud computing myths are growing as the phenomenon gains popularity. Don’t let the promise derail what you need to get done! Myth #1: Cloud computing is an architecture or an infrastructure. Myth #2: Every vendor will have a different cloud. Myth #3: Saa. S is the cloud. Myth #4: Cloud computing is a brand new revolution. Myth #5: All remote computing is cloud computing. Myth #6: The Internet and the Web are the cloud. Myth #7: Everything will be in the cloud. Myth #8: The cloud eliminates private networks.
What Is Cloud Computing? Gartner’s Definition Gartner defines cloud computing as: “ A style of computing where massively scalable IT-enabled capabilities are provided "as a service" to external customers using Internet technologies”. It would be easy to add a refining statement to this definition – to wit, “where the consumers of the services need only care about what the service does for them, not how it is implemented. ” Things to Consider • The cloud versus cloud computing • Private clouds and the public cloud • Lowered barriers to entry, higher risk • Increased scale and elasticity potentially at lower cost • Global class – not just enterprise class
Cloud Computing: Four Perspectives, Four Attributes "All that matters is results; I don't care how it's done" "I don't want to own assets — I want to pay for elastic usage, like a utility" "I want accessibility from anywhere, from any device" "It's about economies of scale, with effective and dynamic sharing" Acquisition Model Service Business Model Pay for usage Access Model Internet Technical Model Scalable, elastic, shareable Cloud Computing: A style of computing where massively scalable IT-enabled capabilities are provided "as a service" to external customers using Internet technologies Global-class industrialization of computing resources and services EC 2 & S 3
How Will Cloud Computing Develop? Technology • Acquisition Model: Richer programmatic SLA interfaces • Business Model: Standardization of usage metrics • Access Model: Improved bandwidth and QOS • Technical Model: Virtualization and automation, RTI, Web services, smart application design Providers Users • Several mega-compute providers will develop (e. g. , watch Akamai, EMC, IBM, Microsoft) – but also many midsized • "Commodity" IT services will go cloud (e. g. , e-mail) • More Saa. S offerings — software business models will be disrupted • Internal IT will also manage dynamic multisourcing from the cloud • Hardware providers will become cloud computing providers, cloud providers become biggest customers • Service providers shift focus from process to service assembly and dynamic change • Barriers to entry will be lower, enabling startups (e. g. , Smug. Mug) • Internal IT will modernize to compete with and leverage the cloud • Some organizations will dramatically shift to the cloud (especially SMBs)
Five Styles of Cloud Computing: An Infrastructure Viewpoint Cloud Software Cloud Processing Cloud Server Software Browser/ access Browser/ access Personal application End-user proc/storage Personal application End-user proc/storage Server application Server proc/storage Cloud Infrastructure Client in the Cloud Server application Server proc/storage
Industrialization of IT: From Custom Software to Assembled Software, to Services Industrial qualities (shared architectures, common skills, standards, reuse and access) Client and server separation makes architecture essential Virtual enterprise emerges, integration becomes essential Distributed client/server applications Monolithic applications Software and OLTP become essential to business Services Solutions as services emerge, pushing solutions as software aside Integrated/ composite applications disruption Mainframe Era Network Era SOA Era Cloud Era Time
What to Expect in the Cloud? IT-Preferred Characteristics (New Instances) Off-Premises Custom • License • Advanced System Control IT-Preferred Characteristics Business Applications 97% Packages • Advanced Differentiation 75% • System Mgmt. Relief • Productivity • Low Cost of Entry • Unique Architecture • Unique Functionality • Unique Applications On-Premises • Assured Isolation (Perception) • Unique Applications • Subscription 2008 2013
Three Reasons Cloud Computing Is Not What You Think Only massively scaled cloud infrastructure providers will succeed – FALSE Midsized providers that can leverage new technologies fast will challenge massively scaled competitors. Mature IT organizations will leverage cloud computing less than less mature – FALSE Enterprises best prepared to leverage cloud will be the most advanced (services-oriented) – or the least (current investments not an issue) Cloud computing is mainly about technology – FALSE The most difficult part of cloud computing is creating a truly services-oriented interface – and delivering to those service requirements – this will take time and invention.
What's Holding Cloud Computing Back? Service Management Technologies & best practices immature Culture Trust, chargeback, sharing Scalability Parallel processing, no problem; sequential processing, different story Connection Only as good as the Internet, unless you pay to "harden" your connection Costs Economies of scale only go so far, unless customer is willing to trade data or advertising views for services High Availability For workloads that are stateless, no problem; for stateful workloads, same issue as enterprises Security & Privacy Weak or nonexistent perimeter, Data provenance, Service/Data Location
Making Sense of the Cloud Alphabet Soup Fixed resources Hardware managed by others Elastic Internet resources Shared applications Cloud Off-Premises Native Cloud Applications Outsourcing Web Hosting Cloud Platform APaa. S AIaa. S Shared application infrastructure (AI) Infrastructure Utility APaa. S – Application Platform as a service size of the cloudlets and overlap shown is not to scale Dedicated Web applications and Web content Programmable or programmatically Dedicated applications accessible resources Saa. S Commodity (industrialized) computing resources Iaa. S – Integration as a Service
Cloud Computing and Cloud/Web Platform • Cloud - Exploiting External, Off-Premise Resources • External mgt/control with internal execution (cache) System Infrastructure Application Information Business Process - New Programming Models - Cloud Enabling Data Centers, Applications & Processes REST Python Map/Reduce DFS • Cloud/Web Platform - Programmatic & programmable - Web interface & architecture - Web services & compositions On Premise or Off Premise Cloud Services Cloud Enablers
Cloud Services — Everything Is a Service • Cloud Service Taxonomy - Cloud/Web Platform 2 System Infrastructure Applications Information Process Ecosystem Management Storage Site Hosting Security Service Provider Information Component Bus. Process Application Ecosystem Mgmt. Service Provider Information Component Bus. Process Application - Provider/Consumer - Builder/Assembler - Individual/Ecosystem Application • Many Cloud/Web Platforms - The consumer's cloud/Web platform is the provider's data center, application or business - Role of services - Revenue generation Compute Application Mgmt. • Approaches • Business Models Cloud/Web Platform 1 Cloud/Web Platform 4 Information Component Bus. Process Composite Assembler Cloud/Web Platform 3
The Service Boundary Determines the Abstraction Level The provider optimizes everything below the service boundary and hides the complexity from the consumer The consumer accesses, configures and/or programmatically extends the service and builds everything needed above the service boundary Solution The Physical Infrastructure inside the cloud matters but indirectly Process Information Cloud Based Services Application Infrastructure System Infrastructure Cloud Enablers Ecosystem Mgmt Solution Enabling Infrastructure Cloud Platform Amazon MTurk EZ Prints Fulfillment Service AWIS Alexa Search Feeds Amazon FPS Force. Com salesforce. com Bungee. Connect Google App Engine Simple DB S 3 IBM Cloud EC 2 Center Services
Building Solutions with Cloud Services (“The Cloud” is a matter of perspective) • Manipulation & Mashup System Infra - Access, program & extend app. , App, Info, info. & process services Process App Infra • Cloud Execution Cloud Mashup - Build and possibly store internally Saa. S Execution — run in cloud Full Cloud - May use mashable services or Platform expose as a mashable service • Full Cloud - Build, store, run, manage in the cloud - May use mashable services or Internal Browser/RIA expose as a mashable service Or Traditional • Hybrid Solution Client - Any combination of internal and Consumer Producer/Manager external for build, store, run, manage
Cloud Computing: A Supply Chain The providers in the cloud create a supply chain — from business process services to component services Services Cloud computing will spur an industry of cloud computing brokers who discover and assemble providers for end users The Internet Service Brokers Software Providers Storage Providers Processing Providers
Infrastructure and Operations: From Silos to Clouds Cloud computing is a natural outcome of next-generation data centers Sprawled Virtualized Automated Component-Orientation Layer-Orientation Service-Orientation Policies • Cloudenabled 2002 -2012 Provisioning Workloads Data Resources Identities Services 2010 -2020 Availability • Service levels and agility up • Hardware costs down, flexibility up Optimization
The Vision for Infrastructure and Operations A real-time infrastructure is: n n n An IT infrastructure shared across customers, business units or applications … …where business policies and service-level agreements drive dynamic and automatic optimization of the IT infrastructure … …thereby reducing costs while increasing agility and quality of service. Efficient use of resources to service policies. Optimization Self-discover, install and integrate. Resources Identities/Security Availability Policies: n IT service definitions n Service agreements n Business priorities Provisioning Workloads/Data Services: n That meet business requirements Avoid, predict and react to failures.
Modernizing Infrastructure: RTI and Cloud Computing Traditional Computing Acquisition Model Business Model Access Model Technical Model RTI, Infrastructure Utility Cloud Computing Buy assets and build technical architecture Buy service Pay for fixed assets and overhead Pay based on usage Internal network or intranet, corporate client Internal network or intranet, any device Internet, any device Single tenant, nonshared, static Scalable, elastic, dynamic, shareable Scalable, elastic, dynamic, multicustomer
Virtualization: Consolidation, Deconsolidation and Cloud Computing Saa. S Application Streaming Applications Grid Computing Infrastructure as a Service Operating Systems Software Appliances Hardware Alternate Delivery Virtualization technologies create layers of abstraction — decoupling interfaces — that enable alternate delivery models
The Virtual Machine Phenomenon: A Major Cloud Enabler 60% VM Penetration of Installed Workloads 61% 40% • Cost, power, space savings • Multicore systems • More important: portability, agility VM Penetration • The VM installed base was 2. 9 M in 2007 – about 7% of the opportunity 20% 7% 0% Driving the Trend 2005 2007 2009 2011 2013 Cloud Impact • VMs will be a major container for cloud computing workloads, enabling dynamic distribution and efficiency Predictions: • The installed base of VMs will grow more than tenfold between 2007 and 2011 • One out of every four x 86 workloads deployed or redeployed during 2008 will be installed in virtual machines • By 2012, the majority of x 86 server workloads will be running in a virtual machine
Virtualization Forces Software Pricing and Licensing Models to Morph Fractional use, consolidated • Especially when consolidated to large servers • Multicore trend exacerbating issue Dynamic change • Old friction: Deployments, upgrades are rare • Now: Deployments are fast, resource allocations change quickly Dynamic movement • Old paradigm: Tie software to serial numbers • Now: Workloads can move dynamically between servers Virtual machine packaging • Software-based appliances, pre-integrated software Enabling cloud computing • Shift to pricing based on usage, workload, users, transactions – away from fixed hardware
Virtualization Enables and Motivates Chargeback: A Foundation for Cloud From Fixed to Variable Usage • Virtualization enables a move to utility pricing • Most businesses are not prepared to deal with IT as a variable expense • Tools and measurement standards needed The Danger of a Frictionless Model • Old - Cap. Ex requirement - Time to respond to requests • New - Cap. Ex asynchronous, smaller - Rapid response (30 X faster) • Chargeback helps provide the friction — helps IT run as a business Cloud Impact? Paying per usage is a fundamental characteristic of cloud computing – organizations that embrace chargeback are better prepared for cloud computing
Software Appliances: On-Premise Saa. S – A Stepping Stone A pure software appliance is a software-only solution that delivers pre-defined service(s) through an application-specific interface, with no accessible operating software. • VM packaging enables prepackaged solutions • • Software-based appliances (SBAs) Virtual software packages (VSPs) • Metadata can store relationships between them, service and operational characteristics (OVF is a start) • Can be managed, updated remotely (by provider) – enabling on-premise Saa. S • Can be dynamically moved – from user to/from provider, between providers • Thins role of OS, creates meta OS
System and Application Infrastructure Services Solution - Hardware and system software (such as OS image, message broker, database) • Application Infrastructure - Database, Integration, Development, BPM technology, Registry/Repository, Content Mgt, Provisioning, etc. - App Platform as a Service (APaa. S) • Based on a multi-tenant architecture for creation of software as a service (Saa. S) solutions • Web influenced Application Control & Execution Models - Map/Reduce, DFS • Such as Google, Hadoop - Other parallel models (such as BOINC) Process Information Application Infrastructure System Infrastructure Cloud Enablers Cloud Platform Ecosystem Mgmt • Virtualized System Infrastructure
Platforms, Application Platforms, SOA and Saa. S What you program SOA Application User-facing App Logic Integration/ Composition Back-end App Logic Data Application Platform Integration Platform Application Platform Data Platform Software Infrastructure Platform Infrastructure What you program to What your platform uses and you should not What your program never sees
An Application Platform Stack as a Cloud Service Users Business Application as a Service Users User Organizations Tenant App. View Saa. S-Enabled Business Application Saa. S-Enabled Application Platform as a Service (APaa. S) App. Dev Application Provider Platform Supplier XTP MT App. Server Metadata Business Data Host System Infrastructure Hardware Grid XTP — Extreme Transaction Processing MT — Multitenancy
Saa. S-Enabled Application Platform: Multi-Tenancy, XTP and Fine-Grained Tracking • Multi-tenancy Users Tenant Saa. S Application Saa. S-enabled Application Platform Saa. S Data Platform Infrastructure Hardware - Multi-tenant execution (isolation and sharing) - Tenant-aware security, monitoring, reporting and management - Tenant customization and user personalization within a tenant - Tenant-aware development (tools and metadata) - Tenant on- and off-ramping - User on- and off-ramping - Application on- and off-ramping and version control • Elasticity (just-in-time on-demand computing resources) • Fine-grained usage tracking, metrics and costing • XTP-grade global class advanced scalability, performance and availability • Integration/composition with other on- and offpremises resources
Choosing the Model of Saa. S-enabling for Business Applications Software On-Premises Enterprise Multi-Tenancy Software as a Service Tenant Application Appl. Application Platform AP AP Infrastructure OS Tenant Appl. AP AP Infrastructure OS OS OS I II IV Isolated tenancy Server/OS virtualization Multi-tenant application platform (APaa. S) Multi-tenant business application Inf. /OS
Application Platform as a Service: Strengths and Challenges § Low cost of entry for ISVs § Elastic scalability for all tenants § Relief of system management burdens for users § Rolling versioning aids extensibility, flexibility and antivirus protection § Full stack resource optimization improves price/performance § Community resource sharing reduces costs § Non-standard programming model (typical) § Data integrity and security in third-party hands § New and unproven, especially in mission-critical deployments § Reduced or no control of Service-Levels and Quality-of -Service § Potentially escalating TCO § Largely Unfamiliar vendors
The Spectrum of IT Options Off-Premises Infrastructure Solutions Application Platforms as a Service *Illustration only, not a complete list of vendors **Some vendors (Magic, Cordys) sell a cloud-enabled application platform, but choose not to host an APaa. S. The classic JEE and. NET application platforms are not cloud-enabled.
Application Platform as a Service Bazaar b b b Multi-tenancy Elasticity Micro-charging XTP Interoperability Development tools Management/ Monitoring Composition support On-premises platform for sale (SEAP)? Minimal to none Yes No Useful, but incomplete No Near complete No No No Yes SEAP = Saa. S-enabled Application Platform XTP = e. Xtreme Transaction Processing
Application and Data/Information Services Sweet Spot for WOA-Centric Cloud Platforms Solution Process - Multi-user or multi-tenant applications as a service - Programmatic access, configuration or extension Information Application Infrastructure System Infrastructure • IFAPs, DOM manipulation, APIs Cloud Platform WOA Influence • Components and Widgets Applications - Service assembly Informtion • Information Feeds - Advertising, news and more • Data Access - Product catalog, pricing data, credit score and more. • Search Cloud Enablers Ecosystem Mgmt • Saa. S & Web Applications Programatic Access Web Influe
Web Technologies and Standards — The Essence of the Cloud/Web Platform • URI/URL Resource Identification Innovators Dilemma • RESTful Service Design Max. mainstream • Data-Centric Resource Manipulation - HTTP/XML, Json, DOM, Ajax • XML Standards - SIP, industry models (such as Accord) Functionality - Minimal, generic verbs m Co s rise ice erp Ent erv d. S ign e s as De d-B u tric en Clo -C eb Min mainstream W • Syndication (RSS/Atom) • Semantic Hypertext & Semantic Web ting pu Time 2008 2013
Web Technologies and Standards Influencing Application Development & SOA • New Clients and Access Models Innovators Dilemma - Browsers, Ajax and RIA Max. mainstream • Web-Oriented Architecture - URI/URL Resource Identification - RESTful Service Design - Minimal, generic verbs - Data-Centric Resource Manipulation - HTTP/XML, Json, DOM, Ajax Functionality - Syndication (RSS/Atom) is rpr e Ent l ba lo G • Web Standards g utin p om e. C b. C We s mp o las C Min mainstream Time - SIP, industry models (such as Accord) 2008 g n uti 2013
Business Process Services Web-Enabled B 2 B and B 2 C Commerce Solution - Sites, applications and widgets with action • Business Service Execution - Inputs: Information, event/trigger, dashboard selection, SLA setting - Process: human or automated action in the "real world" - Outcomes: Information/content, events/triggers, physical product • Proxy actions: Outcome delivered to a third party • Existing and New Players - Existing: ADP, Fed-X, UPS - New: e-commerce, Web advertising, BPO, everyone else Process Information Application Infrastructure System Infrastructure Cloud Enablers Cloud Platform Ecosystem Mgmt • Business Actions
Key Trends in Cloud Computing The Trends • Cloud computing provides global-class industrialization of computing resources and is focused on external service delivery • Everything is a service … But not everything is purchased as a service • Cloud/Web platforms provide programmatic access to Web-based services which act as a foundation to create compositions • Internal data-center-based cloud/Web platforms are as important as external cloud -based cloud/Web platforms • One company's data center or business process may be another company's cloudbased service • Virtual machines do more than consolidate – they change computing paradigms, cultures, etc. • The biggest challenge in cloud computing is rich service orientation – defining requirements and meeting them. • Providers need not themselves be massive but must be able to provide the necessary scalability to meet the expanding and contracting needs of consumers on demand economically.
What we are telling the Enterprise CIOs and enterprise architects should… ü Today (Monday Morning) - Embrace WOA principles, standards and best practices to create an internal cloud/Web platform for applications and information services - Direct emerging technology teams to evaluate cloud computing and cloud/Web platform technologies, and explore various adoption scenarios ü Near Future (The Next 12 Months) - Demand that your technology providers explain how cloud computing will affect their offering and pricing strategies - Experiment with external cloud based services, establish governance models and exploit virtualization and RTI techniques to build cloud infrastructure internally. ü Longer Term (Three to Five Years) - Employ ecosystems management services and techniques to ensure quality of service across cloud vendors and cloud platforms. - Determine if, when and how the enterprise should expose its information, application and/or processes to others as cloud services