Commodity computing: Difference between revisions

Revision as of 06:25, 6 March 2010

Revision as of 06:36, 6 March 2010

Line 1:

{{Unreferenced|date=August 2007}}

'''Commodity computing''' is computing done in commodity computers as opposed to supermicrocomputers or boutique computers. Commodity computers are [[computer system]]s manufactured by multiple vendors, incorporating components based on [[open standard]]s. Such systems are said to be based on [[commodity]] components since the standardization process promotes lower costs and less differentiation among vendor's products. A governing principle of commodity computing is that its better to have more lower performance, lower cost hardware working in parallel [[Scalar computing]] (eg [[AMD]] x86 [[CISC]]<ref>http://www.computerworld.com/s/article/9154518/IBM_HP_servers_won_t_stop_x86_onslaught_on_Unix</ref> ) than it is to have less but more expensive hardware <ref>http://research.google.com/pubs/DistributedSystemsandParallelComputing.html</ref> (eg IBM [[POWER7]]<ref>ftp://ftp.software.ibm.com/common/ssi/pm/rg/n/poo03017usen/POO03017USEN.PDF</ref> [[RISC]] ). At some point the number of discrete systems in a cluster or cloud will be greater than the [[MTBF]] for any hardware platform, no matter how reliable, fault tolerance must be built into the controlling software<ref>http://www.morganclaypool.com/doi/abs/10.2200/S00193ED1V01Y200905CAC006</ref><ref>http://insidehpc.com/2008/06/02/google-fellow-sheds-some-light-on-infrastructure-robustness-in-face-of-failure</ref>. Purchases should be optimized on cost per unit of performance, not just absolute performance per CPU at any cost.

== History ==

Line 69:

*[http://labs.google.com/papers/mapreduce-osdi04-slides/index-auto-0021.html Fault tolerance Handled via re-execution]

*[http://hadoop.apache.org/ HADOOP]

*[[http://google-services.blogspot.com/2006/07/google-machine.html Google Commodity computing models]

[[Category:Computing platforms]]