
1d1117056ebec31e41e212f51e6de4c6.ppt
- Количество слайдов: 30
Enabling continuous quality assurance and optimization in future enterprise cloud service brokers What is Cloud Computing? Anthony J H Simons (University of Sheffield) Call: FP 7 -ICT-2011 -8 Grant agreement no: 318392 cloud. eu www. broker-
Overview • What the cloud is and isn't • Defining characteristics • Cloud service models – Infrastructure-as-a-Service – Platform-as-a-Service – Software-as-a-Service • Cloud service ecosystems – stores and marketplaces – customisation and brokerage 19/03/2018 Anthony J H Simons, University of Sheffield 2
What is the Cloud? • Enabling technology – web, datacentres • What the cloud isn't – online working – grid computing – service-oriented © Estate of Mrs J C Robinson / Pollinger Ltd / Mary Evans Picture Library 19/03/2018 Anthony J H Simons, University of Sheffield 3
Cloud = Online? • Mail/Chat – AOL, Hotmail, Yahoo, Gmail, IM, Skype • Data – Dropbox, Flickr, One. Drive, i. Cloud • Sales – Pay. Pal, e. Bay, i. Tunes, App. Store, Ocado • Social – Facebook, Twitter, Linked. In, You. Tube, Tumblr, Pinterest, Instagram, Snapchat • Virtual – Second Life, World of Warcraft, MMORPGs 19/03/2018 Anthony J H Simons, University of Sheffield 4
Cloud = Integration? • Google – Office: Mail, Calendar, Contacts – CSCW: Groups, Sites, Drive – Other: Maps, Translate, Blogger • Apple – Devices: Mac, i. Pad, i. Phone – Services: i. Tunes, App. Store, i. Cloud • Microsoft – Sign-in: Windows 8. x, One. Drive, Skype 19/03/2018 Anthony J H Simons, University of Sheffield 5
Cloud = Distributed? • Distributed Computing – – centralised datacentres (benefit of scale) physically distributed nodes (attack-proof) secure networking (encrypted) data replication, recovery (RAID standard) • Grid Computing – – massively parallel, networked compute engines Hadoop style architecture (YARN, node-aware DFS) Map. Reduce style processing (PDP on a cluster) solves SIMD style number crunching on Big Data 19/03/2018 Anthony J H Simons, University of Sheffield 6
Cloud = SOA? • Service-Oriented Architecture – web-enabled business services • online wrappers around older ERP systems • atomic services, "componentized" applications – standard communication protocols • Web Services Description Language (WSDL) • Simple Object Access Protocol (SOAP) – service discovery and integration • registry and repository systems (UDDI – old) • service composition – orchestration, choreography • Business Process Execution Language (BPEL) 19/03/2018 Anthony J H Simons, University of Sheffield 7
What is the Cloud? • What the cloud is – virtualised, shared computing resources – elastic computing, arbitrary scaling – utility consumption model © Estate of Mrs J C Robinson / Pollinger Ltd / Mary Evans Picture Library 19/03/2018 Anthony J H Simons, University of Sheffield 8
Origin of "the Cloud" • Telecoms – cloud symbol = "the rest of the Internet" – computer network diagrams (1994) • boundary of LAN, router connects to cloud • Compaq – internal memo on "cloud computing" (1996) • future of internet-based business • Amazon – launch of Elastic Compute Cloud (EC 2) (2006) • popularised use of the term "the cloud" 19/03/2018 Anthony J H Simons, University of Sheffield 9
Milestones • Salesforce (1999) – pioneered HR, CRM, sales software via the web • Amazon (2002) – Amazon Web Services (AWS) infrastructure – storage, computation, marketplace for workers • Amazon (2006) – Elastic Compute Cloud (EC 2) & Simple Storage Service (S 3) – clients rent computers, run own applications • Google (2009) – Google Apps, mail, sites, docs (=drive), maps … 19/03/2018 Anthony J H Simons, University of Sheffield 10
NIST Definitions • On-demand – obtain servers, CPUs, O/S, storage as needed • Network access – accessible by phones, tablets, laptops, PCs, … • Resource pooling – multi-tenancy, resources reassigned on demand • Elasticity – rapid scaling in/out commensurate with demand • Measured – monitored bandwidth, storage; metered charging 19/03/2018 Anthony J H Simons, University of Sheffield 11
Kinds of Cloud • Private Cloud – internal, hosted by company for its own needs • Public Cloud – Amazon, Google, Rackspace, etc. open to anyone • Hybrid Cloud – private + public, especially when "cloud bursting" • Community Cloud – pooled infrastructure for community with similar goals • Federated Cloud – any pooling of multiple public and private clouds 19/03/2018 Anthony J H Simons, University of Sheffield 12
Cloud Service Models • Infrastructure as a Service (Iaa. S) • Platform as a Service (Paa. S) • Software as a Service (Saa. S) 19/03/2018 Anthony J H Simons, University of Sheffield 13
The Cloud Stack Client web browser Client mobile phone Saa. S email, office, CRM, HR, ERP, chat, maps, desktop apps… Paa. S virtual desktop, sign-in, DB engine, webserver, app repository, API tools… Iaa. S Client thin client PC 19/03/2018 servers, CPUs, network, storage, load balancers, virtual machines, O/S… Anthony J H Simons, University of Sheffield 14
Infrastructure as a Service • Commercial benefits – avoid owning and maintaining hardware (SMEs) – or pool usage of all owned infrastructure (IBM) • Virtualisation of hardware – needs a hypervisor, e. g. Xen, Virtual. Box, Hyper-V – allocates 1. . n virtual machines (VMs, "instances") – uploads complete O/S image (Ubuntu, Windows) • Elastic consumption – physical nodes allocated on-demand – load balancing system maximises usage 19/03/2018 Anthony J H Simons, University of Sheffield 15
Elastic Scaling • Vertical scaling – concurrent multicore – scale up: add more CPU cores and RAM – scale down: free up excess CPUs and RAM • Horizontal scaling – distributed VMs – scale out: spool up more VMs from a ready pool – scale in: turn off unused VMs, return to pool • Different kinds of hypervisor – "native" type 1: runs on bare metal, offers guest OS to clients – "hosted" type 2: sits on server OS, offers guest OS to clients 19/03/2018 Anthony J H Simons, University of Sheffield 16
e. g. Amazon Web Services • Computing services – Amazon Elastic Compute Cloud (EC 2) – clients boot 1. . n VMs through web interface – instances contain whatever software desired • Storage services – Amazon Relational Data Service (RDS) – full relational DB capability, normalised tables – Amazon Simple Storage Service (S 3) – block storage, like a simple filesystem – Amazon Glacier – long-life cold storage, cheap to store and keep, expensive to retrieve 19/03/2018 Anthony J H Simons, University of Sheffield 17
Platform as a Service • Commercial Benefits – avoid maintaining O/S platform and standard apps – application Paa. S: build apps out of other apps – integration Paa. S: wrap legacy ERP systems • Platform Features – – – web hosting automatic load balancing system development kit (SDK – app builder) reusable services/components (single sign-in) marketing and distribution (app store/marketplace) single point of billing (manage revenues to ISVs) 19/03/2018 Anthony J H Simons, University of Sheffield 18
Kinds of Platform • Generic open platforms – hard work – offer basic OS, developer kit, storage services – add your webserver, Tomcat, JBoss, applications – e. g. Open. Shift, Cloud. Foundry • Open platforms with SDK – moderate – offer basic services, sign-in, mail, MQs, no. SQL DBs – APIs to create apps in Java, Python, PHP, Go, … – e. g. Google App. Engine, Heroku, Amazon Elastic Beanstalk • Closed virtual desktops – "vendor lock-in" – web-based, or "rich client" virtual desktop environment – select and deploy ready-made apps using point and click – e. g. Zoho Creator, Salesforce 1 19/03/2018 Anthony J H Simons, University of Sheffield 19
e. g. Google App. Engine • Runtime support – APIs for Python, Java, Go and PHP – and any JVM-compatible Groovy, Scala, Clojure • Google-style apps – mail, calendar, contacts, drive, sites, maps – easier to write automatically-scalable apps • but range of apps limited to fit the infrastructure • Data storage – uses GQL, not SQL, to avoid joins spanning machines • special reference property for 1: M, M: N relations 19/03/2018 Anthony J H Simons, University of Sheffield 20
Software as a Service • Commercial Benefits – avoid purchasing shrink-wrapped software (and recurrent costs of upgrading to v 2, v 3, …) – rent latest-version products, on-demand as needed • Software Features – – generic packages for mail, calendar, contacts business packages for HR, CRM, accountancy wrappers for banking, taxation, legacy ERP custom extensions (provided by ISVs) 19/03/2018 Anthony J H Simons, University of Sheffield 21
Kinds of Service API • SOA: standard protocols – traffic uses WSDL interfaces; SOAP message data – thin client, all processing server-side – slow but standard, modular and secure • REST: popular convention – traffic uses HTTP GET, POST; JSON or XML data – thin client, all processing server-side – medium, modular and secure, but bespoke • Rich Client: vendor specific – traffic uses AJAX "data trickle"; bespoke CRUD on server – rich client desktop, full MVC for business processing – fast, but risks malware penetration on client 19/03/2018 Anthony J H Simons, University of Sheffield 22
e. g. Salesforce 1 • Salesforce. com – best known CRM/sales dashboard online – used to be called Force. com, … • Salesforce 1 platform – – – – Sales Cloud – field sales on mobile devices Service Cloud – build customer loyalty Marketing Cloud – create campaigns Community Cloud – connect partners, customers Data. com – all contact and account data Work. com – team performance, metrics, goals Desk. com – customer service, faster inbox handling 19/03/2018 Anthony J H Simons, University of Sheffield 23
Trading in the Cloud • Cloud App Stores – – sense of a regulated shop vetted and "taxed" by platform Apple App Store (i. OS/Mac) Salesforce App. Exchange • Cloud Marketplaces – – sense of an unregulated market "open trading" (nearly) Google Play (Android Market) Amazon AWS Marketplace 19/03/2018 Anthony J H Simons, University of Sheffield 24
Emerging Ecosystems • Platform provider – some in-house apps – outsources some apps to 3 rd party ISVs • Software provider – registers as a partner – creates app plugins, extensions • Consumer – benefits from bundle of customised services 19/03/2018 Anthony J H Simons, University of Sheffield 25
e. g. Heroku, Amazon & … 3 rd Party Saa. S Providers Heroku Paa. S virtual desktop, sign-in, app store… mail/SMS Postmark message queues Cloud. AMQP DB engines AWS Iaa. S offerings Postgres Mail. Chimp Rabbit. MQ Redis Pub. Nub Mongo Amazon Web Services Iaa. S VMs, filestore, load balancer RDS 19/03/2018 Blower S 3 Glacier Anthony J H Simons, University of Sheffield 26
NIST Cloud Roles • Cloud Consumer – user of services, end-user, or service composer • Cloud Provider – provider of services, possibly composed of others • Cloud Broker – intermediary, added-value, facilitator for ecosystems • Cloud Auditor – security auditing, legal compliance to local laws • Cloud Carrier – network provider, mobile carrier 19/03/2018 Anthony J H Simons, University of Sheffield 27
Service Brokerage • Discovery – host or advertise repositories of services and search engine • Integration – build service wrappers around legacy ERP systems • Aggregation – offer bundled applications from sets of services • Customization – build or host adapted or extended versions of services • Optimization – monitor value or performance, provide arbitrage, choice of offering • Continuous QA – provide governance of service lifecycle, certification, testing • Failure prevention – monitor trends, scale out, switch providers, recover from failure 19/03/2018 Anthony J H Simons, University of Sheffield 28
Questions • Acknowledgement FP 7 -ICT-2011 -8 grant agreement no: 318392 • Project website www. broker-cloud. eu 19/03/2018 Anthony J H Simons, University of Sheffield 29
References • • • NIST, The NIST Cloud Computing Reference Architecture, Special Publication 500 -292, 2011. D Kourtesis, I Paraskakis, A Simons: Policy-driven governance in cloud application platforms: an ontology-driven approach, Technical Report, SEERC, 2012. Broker@Cloud D 20. 1 State of the art and research baseline, http: www. broker-cloud. eu, 2013. Broker@Cloud D 20. 3 Requirements for continuous quality assurance and optimisation in cloud brokerage, http: www. broker-cloud. eu, 2013. F Gonidis, I Paraskakis, A Simons, Leveraging platform basic services in cloud application platforms for the development of cloud applications, Proc. Cloud. Com, 2014. 19/03/2018 Anthony J H Simons, University of Sheffield 30
1d1117056ebec31e41e212f51e6de4c6.ppt