You are on page 1of 26

$istribute% Systems an% GRI$ Com&utin'

Author: Reisel Gon !le "#re


University of the Informatic Sciences Havana- Cuba 2012

Sumary

1( Com&uter from the be'innin's( 2( Challen'es( )( Hi'h "erformance Com&utin'( *( $istribute% Systems(

Com&uter from the be'innin's

History

The Legendary Chinese Abaco

History

The first true digital computer was designed by the English mathematician Charles Babbage - Difference Engine

Today

ith the in!ention of the microchip computers became more powerful and essential to the de!elopment of science and life in general""""

Actually e+ist com&utin' &o,er to solve ma-or &roblems an% challen'es of science(((

Challenges

Large Hadron Collider (LHC)

This pro#ect is e$pected to generate more than 37 terabytes of data every day... (1 TB = 1024 GB) This information must be stored% analy&ed and processed by High Performance Computing (HPC) systems"""

Hi'h "erformance Com&utin'

High Per orman!e Com"uting

The high-performance computing is a !ery important tool in de!eloping computer simulations to comple$ problems" Based on computer technologies such as clusters% supercomputers% or using parallel computing"

HPC # $""li!ation $reas

Clustering

Cluster computers% Di!ision of +hysical Chemistry )C,-B* that stores and protects information from genes and proteins"

Cluster computers of the 'ational Aeronautics and (pace Administration )'A(A*% go!ernment agency responsible for space programs"

%ost Po&er ul Su"er!om"uter

HPC # %ain 'isad(antage

The main issue on .C+ is the cost" Build or buy a supercomputer is !ery e$pensi!e" The / Computer cost o!er $1.25 billion US dollars"

$istribute% Systems an% GRI$ Com&utin'

Grid Com"uting

Distributed computing paradigm in which all the resources of any number of computers are subsumed to be treated as a Single irtual Supercomputer"

)*+eti(es

Distributed computing and -0,D Computing is designed to sol!e problems too big for any supercomputer or cluster"

,eatures

Distributed systems must be scalable! transparent and fault"tolerant"

Terms

Scalabilit#$ refers to the variability of the size of a distributed system. %ransparenc#$ refers to the invisibility of the use of multiple processors and remote access. &ault %olerance$ if a system component breaks another component must be able to replace.

$d(antages
1" +ro!ides a mechanism for transparent collaboration between dispersed computer groups" 2" Enables the operation of large-scale applications" 3" +ro!ides access to distributed resources from our computers" 4" All these ob#ecti!es and benefits are included in the idea of 5e-(cience5

%ain ad(antage

,n a cluster all computers are in the same place% connected by a LA'% those are dedicated computers and must ha!e the same hardware" ,n contrast% in the -0,D computers can be distributed around the world"

%a+or "ro+e!ts
1" (eti6home7 radio signal processing to search for e!idence of e$traterrestrial intelligence" 2" 8olding6home7 simulating protein folding% primarily using molecular dynamics techni9ue" 3" 8ightA,D(6home7 using biomedical simulation techni9ues to loo: for ways to cure or pre!ent the spread of A,D( and .,;" 4" The -rid7 process and store the information generated by the 5Large .adron Collider"5

-e eren!es
http7<<lcg"web"cern"ch<LC-< http7<<www-fp"mcs"anl"go!<=foster<Articles< hat,sThe-rid"pdf http7<<www"redboo:s"ibm"com<abstracts<sg24>>?@"html http7<<www"redboo:s"ibm"com<abstracts<sg24>AAB"htmlCDpen http7<<www"top?@@"org< ,an 8oster% Carl /esselman )1EEE*"The Grid: Blueprint for a New omputin! "nfrastructure" Forgan /aufmann +ublishers" ,(B'" 8ran Berman% Anthony G"-" .ey% -eoffrey 8o$ )2@@3*"Grid omputin!: #akin! The Global "nfrastructure a $eality " iley" ,(B'"

Questions

$istribute% Systems an% GRI$ Com&utin'


Author: Reisel Gon !le "#re
University of the Informatic Sciences Havana- Cuba 2012

You might also like