Cloud and grid computing concepts

Cloud computing

Cloud computing delivers computing resources (servers, storage, databases, software, networking) as on-demand services over the internet, with pay-per-use billing. Users do not own or manage the physical hardware.

• Service models: IaaS (virtual machines, storage; e.g. AWS EC2), PaaS (platform to build apps; e.g. Google App Engine), SaaS (ready software; e.g. Gmail).
• Deployment models: public, private, hybrid, community.
• Characteristics: on-demand self-service, broad network access, resource pooling (virtualisation), rapid elasticity, measured service.
• Benefits: low upfront cost, scalability, accessibility, reliability. Limits: security and privacy, dependence on internet and vendor lock-in.

Grid computing

Grid computing links geographically distributed, heterogeneous computers from different organisations into one virtual supercomputer to solve very large problems. Idle CPU power and storage are shared; a middleware (such as Globus Toolkit) schedules jobs, manages resources, security and data transfer.

• Used for scientific problems: weather prediction, particle physics (CERN LHC grid), SETI@home, drug discovery.
• Computers are loosely coupled, and a large job is split into small independent tasks.

Difference

 Users --> [Internet] --> Cloud data centre (VMs, storage)
 Job --> [Grid middleware] --> PC1, PC2, cluster, PCn
                                (results merged)