80 likes | 94 Views
This course provides an overview of Grid Computing, explaining its differences from parallel and distributed computing. It explores the need for a collaborative and challenging environment for various applications and discusses the research challenges and opportunities in the field.
E N D
Topics in Grid ComputingOrientation Sathish Vadhiyar
Grid?? • What is a Grid? A bunch of machines put together for problem solving • Why need a bunch of machines? • For solving quickly • For solving larger sized problems • To improve the quality of the solutions • Not all components for solving a problem reside on a single machine • Many more
What’s new • How is Grid computing different from the earlier parallel computing, distributed computing etc.? • Shared resources with multiple ownerships • So? Many of the assumptions you make on traditional resources go away. Multiple system administrators, multiple usage policies Load patterns in the resources can vary Resources can be arbitrarily brought down
New Research Challenges • Interfacing/Integrating multiple research techniques • Multiple security protocols • Multiple job spawning protocols • Multiple user interfaces • Multiple scheduling policies • ….. • Scheduling and rescheduling • Fault tolerance and migration • Identifying and developing new applications on the Grid paradigm
But, why do we need such a challenging environment? • For ever-greedy applications • For long running applications • For multi-component applications • For parameter sweep applications • For collaborative computing • For remote resource usage
Why the term Grid • Analogy to the electrical power grid • 20,000 generators, billions of outlets, different ownerships • Various appliances, different kinds of users • Ease of use
About the Course • Course web page:http://www.serc.iisc.ernet.in/~vss/courses/GC2008 • Course Details • Research-oriented course • Timings • Grading scheme • Syllabus • Other rules of the game.. • Registration…
Definitions Relevant to the Course • Seamless access • Real scientific (parallel) applications • Automatically acquiring or shrinking resources • Linking disparate software resources • Something that seamlessly facilitates something (computing or whatever) either not possible or possible with much severe constraints on local (owned, 1-site) resources • Resources have to be heterogeneous in some way or other. • Has to involve at least 2 different sites to realistically present network heterogeneity