Cloud Computing Dr Elise de Doncker CS 6260

Cloud Computing Dr. Elise de Doncker CS 6260 Yazeed K. Almarshoud

Roadmap l l l l Introduction Parallel vs. Distributed l Grid computing structure l Flynn’s Taxonomy Cloud vs. Grid Cloud Computing l Possibilities l Some Characteristics of Cloud Computing l Saa. S and Cloud Computing l Supercomputing & Cloud Computing Clouds Examples Conclusions References

Introduction l During the good economic times, enterprises do huge investment in Information Technology (IT) infrastructure to achieve faster and reliable response to users’ queries. l The concept of parallel computing & distributing systems widely used and enhanced in many related environments (. i. e Grids) What is exactly the difference when we say Parallel or Distributed?

Parallel vs. Distributed l Parallel computing generally means: l l l Vector processing of data Multiple CPUs in a single computer Distributed computing generally means: l Multiple CPUs across many computers

Flynn’s Taxonomy Multiple (MD) Single (SD) Data Instructions Single (SI) Multiple (MI) SISD Singlethreaded process SIMD Vector Processing MISD Pipeline architecture MIMD Multithreaded Programming

SISD Processor D D Instructions D D D

SIMD Processor D 0 D 0 D 1 D 1 D 2 D 2 D 3 D 3 D 4 D 4 … … … … Dn Dn Instructions

MIMD Processor D D D D D Instructions Processor D D Instructions

Parallel vs. Distributed Processor D D D D D Network connection for data transfer D Instructions Shared Memory Processor D D Instructions Parallel: Multiple CPUs within a shared memory machine Distributed: Multiple machines with own memory connected over a network

Divide and Conquer “Work” Partition w 1 w 2 w 3 “worker” r 1 r 2 r 3 “Result” Combine

Grid Computing Structure (big picture)

Cloud computing l “Cloud computing is a computing paradigm shift where computing is moved away from personal computers or an individual application server to a “cloud” of computers. Users of the cloud only need to be concerned with the computing service being asked for, as the underlying details of how it is achieved are hidden. This method of distributed computing is done through pooling all computer resources together and being managed by software rather than a human. “

Cloud vs. Grid l l l Cloud Computing is an infrastructure that virtualizes hardware and software resources Grid Computing are patterns, tools and frameworks to distribute computing or data A cloud can be the platform to run a computing or data grid

Cloud Computing l l Cloud computing is a novel platform for computing and storage. Cloud computing provisions and configures servers as needed. It allows for more efficient use of the enterprise resources and applications. It introduces accountability and streamlines computing needs of an enterprise.

Possibilities l l It is possible to consolidate all the needs of an organization in a systematic and accountable fashion. It is possible to procure computing related resources similar to how you rent a place for living. For example, l you can buy storage on demand from amazon. com in a service it offers called the “S 3” l You can buy computation service from amazon. com in its “elastic cloud computing” service (EC 2) Usage example: You are in charge of IT in a local company. You have an immediate need for backing up entire set up for a short period of time as a mock up for disaster recovery. What would you do?

What is driving Cloud Computing • Technology advances that support massive scalability & accessibility • Emergence of data intensive applications & new types of workloads àLarge scale information processing, i. e. parallel computing using Hadoop àWeb 2. 0 rich media interactions àLight weight run anywhere web apps Skyrocketing costs of power, space, maintenance, etc. Explosion of data intensive applications on the Internet Advances in multi-core computer architecture Fast growth of connected mobile devices Growth of Web 2. 0 enabled PCs, TVs, etc.

Industry Trends Leading to Cloud Computing 2008 2000 1998 Grid Computing Solving large problems with parallel computing Made mainstream by Globus Alliance Utility Computing Offering computing resources as a metered service Introduced in late 1990 s Software as a Service • Network-based subscriptions to applications • Gained momentum in 2001 Cloud Computing • Next-Generation Internet computing • Next-Generation Data Centers

Some Characteristics of Cloud Computing l Virtual – Physical location and underlying infrastructure details are transparent to users l Scalable – Able to break complex workloads into pieces to be served across an incrementally expandable infrastructure l Efficient – Services Oriented Architecture for dynamic provisioning of shared compute resources l Flexible – Can serve a variety of workload types – both consumer and commercial

Cloud Computing in the New Enterprise Data Center Workload Solution Patterns Software Development Technology Incubation Innovation Enablement Deploys development tools for immediate use Reduces time to launch new offerings Expands sources of innovation, increases competitiveness Large Scale Information Processing Optimizes emerging Internet scale workloads Cloud Computing Management Services Self-service Admin Portal Workload Pattern Templates Workload Management Administration Workflows Provisioning SLA and Capacity Planning Monitoring Virtualized Physical Servers (Ensembles) i. Data. Plex, Blade. Center, System x, System p, System z

Why Cloud Computing? l l l Pay per use Instant Scalability Security Reliability APIs [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Case Study of a Cloud Deployment 100% New Development Software Costs Power Costs Current IT Spend Liberated funding for new development, transformation investment or direct saving Strategic Change Capacity Deployment (1 -time) Labor Costs (Operations and Maintenance) Software Costs Power Costs (88. 8%) Hardware Costs (annualized) Labor Costs ( - 80. 7%) Hardware Costs ( - 88. 7%) Note: 3 -Year Depreciation Period with 10% Discount Rate Hardware, labor & power savings reduced annual cost of operation by 83. 8%

“Cloud Computing” Defined “as a Service” types l l l Everything as a service (Eaa. S or Xaa. S) Communication as a service (Caa. S) Infrastructure as a service (Iaa. S) Monitoring as a service (Maa. S) Software as a service (Saa. S – includes Application Service Provider (ASP) services) l Platform as a service (Paa. S) [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Software as a Service Paa. S Platform as a Service Iaa. S Infrastructure as a Service [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Software as a Service [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Software delivery model l Increasingly popular with SMEs No hardware or software to manage Service delivered through a browser [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Advantages l l l Pay per use Instant Scalability Security Reliability APIs [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Examples l l CRM Financial Planning Human Resources Word processing Commercial Services: l l Salesforce. com emailcloud [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Paa. S Platform as a Service [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Platform delivery model l Paa. S l l Platforms are built upon Infrastructure, which is expensive Estimating demand is not a science! Platform management is not fun! [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Popular services l Paa. S l l Storage Database Scalability [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Advantages l Paa. S l l Pay per use Instant Scalability Security Reliability APIs [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Examples l Paa. S l l Google App Engine Mosso AWS: S 3 [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Iaa. S Infrastructure as a Service [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Computer infrastructure delivery model Access to infrastructure stack: l l Iaa. S l l Full OS access Firewalls Routers Load balancing [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Advantages l l Iaa. S l Pay per use Instant Scalability Security Reliability APIs [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Examples l l Flexiscale AWS: EC 2 Iaa. S [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Software as a Service Paa. S Platform as a Service Iaa. S Infrastructure as a Service [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Common Factors l Paa. S l l Pay per use Instant Scalability Security Reliability APIs Iaa. S [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Advantages l Paa. S l l Iaa. S l Lower cost of ownership Reduce infrastructure management responsibility Allow for unexpected resource loads Faster application rollout [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Saa. S Cloud Economics l Paa. S l l Iaa. S l Multi-tenented Virtualisation lowers costs by increasing utilisation Economies of scale afforded by technology Automated update policy [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Supercomputing & Cloud Computing l l Two macro strategies dominate large-scale (intentional) computing infrastructures Supercomputing type Structures l l l Large-scale integrated coherent systems Managed for high utilization and efficiency Emerging cloud type Structures l l Large-scale loosely coupled, lightly integrated Managed for availability, throughput, reliability

How should we think about the cloud opportunities? l l l Virtual zoo of systems? Replacements for Clusters? Extensions to existing systems and infrastructure? l l l Surge capacity? Edge datasystems? Opportunity to go “hardwareless” when designing new systems and services?

The Virtual Zoo l Access to a diverse image library provides an inexpensive mechanism to test applications and services on a variety of OS configurations without having to build all of them. l l l Leverages virtualization and community images Leverages “cloud” when scale is important Using cloud for scalability testing could be interesting when you have servers you want to stress and test, but limited time and resources l l Creating hundreds of running instances is relatively easy and could be done by a few people in less than a day Automation of the scalability testing could be easily accomplished

As Replacements for Clusters? l There have been several experiments creating virtual clusters in EC 2 and probably in other environments as well [Peter Skomoroch, et al]. l These “soft” clusters are interesting, constructed on demand then torn down with the application run is complete. l It might be possible to integrate virtual clusters into existing Linux cluster queues such that jobs that are queued for a physical cluster could be dispatched to a local cluster or a cloud based virtual cluster for execution. l l l In fact for throughput jobs this might be even more effective. Local facilities that start supporting image based scheduling services would lead in this transition (i. e. you submit your job as one or more images rather than scripts or executables) Cloud hosting for clusters provides one easy way to implement cycle banking since each application determines their own operation environment and overheads are relatively low l l This would ideally be implemented as a distributed resource if physical ownership was important Virtual ownership would make it much easier and robust to implement

Seamless extensions l l Like in the previous example seamlessly extending an existing queue could be a one way to integrate clouds with existing services and systems. But we can imagine others. How about using the cloud as a giant impedance matcher for geographically distributed systems of large-scale sensors and tightly coupled data analysis environments? The idea is simple. [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Surge Capacity l Power companies have peakers. l l Typically natural gas powered turbines used during times of peak demand for power. Clouds can be used for surge capacity for groups that have variable demands for access to compute cycles or server/service cycles

Sensor + Cloud + Supercomputer = Next Generation Simulations l Imagine thousands (or millions) of distributed sensors deployed over the globe each generating data in some asynchronous fashion. l Each sensor updates data structures in the cloud via local internet connections. The cloud is ubiquitous, secure enough, reliable etc. and scales to the size of the sensor network and acts as an impedance matcher. l Periodically harvesting processes (in the cloud say) wake up and organize the datasets into a fashion that they can be downloaded coherently to a supercomputer for data assimilation to a large-scale parallel simulation.

Going Hardwareless l l Need: 24 x 7 access to flexibly configured hardware, scalable data infrastructure, and customized operating environment 1000 cores x. 10 hour x 8760 hours/year x 3 years = $2. 6 M 1000 cores x $390/core + 3 x $43, 800 power + 3 x 200 K + 3 x 100 K = $1. 4 M In my example if cluster utilization is < 53% then it is cheaper to go “hardwareless” at current retail prices [An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney]

Clouds Examples l Amazon. com l l l Amazon Simple Storage Service (Amazon S 3). Amazon Elastic Compute Cloud (Amazon EC 2) Hadoop (Map/Reduce) l Large scale information processing, i. e. parallel computing

Conclusions l l l l The emerging concept of the cloud is pretty cool. The existing available “retail” models are hugely empowering, since they require only a credit card to get going. Ease of use is being tackled, a market is developing for images and value added services. Clouds feel like the next thing that will have traction and will enable hardwareless ventures. Scientific applications will not drive clouds, but will benefit from their widespread adoption. It is a disruptive technology in many ways and the university/agency shift will take some time, hence private sector will likely get significantly ahead. Many groups should be experimenting and it really is pretty cheap to gain the critical experience to figure out interesting things to try.

References l http: //en. wikipedia. org/wiki/Cloud_computing l Includes references to Amazon, Apple, Dell, Enomalism, Globus, Google, IBM, Knowledge. Tree. Live, Nature, New York Times, Zimdesk l Others like Microsoft Windows Live Skydrive important l An Introduction to Saa. S and Cloud Computing presentation By Ross Cooney l http: //en. wikipedia. org/wiki/Amazon_Elastic_Compute_Cloud http: //uc. princeton. edu/main/index. php? option=com_content&task=view& id=2589&Itemid=1 Policy Issues http: //www. cra. org/ccc/home. article. bigdata. html l Hadoop (Map. Reduce) and “Data Intensive Computing” l See Data intensive computing minitrack at HICSS-42 January 2009 l l http: //ianfoster. typepad. com/blog/2008/01/theres-grid-in. html l OGF Thought Leadership blog OGF 22 talks by Charlie Catlett and Irving Wladawsky-Berger

Presentation Question: What are the two macro strategies dominate large-scale (intentional) computing infrastructures? Explain. l. Supercomputing type Structures l. Large-scale integrated coherent systems l. Managed for high utilization and efficiency l. Emerging cloud type Structures l. Large-scale loosely coupled, lightly integrated l. Managed for availability, throughput, reliability