Advanced

Google and IBM push cluster computing on colleges

Posted by Scott_Ruecker on · 4 comments
Reg Developer; By Austin Modine 		Mail this story
Print this story

Google and IBM want their future employees to have large-scale cluster computing chops, so they're investing several million to get them while their young. The companies are teaming up to promote the study in academia. Their ambition is to lower the cost and logistics of collegial research on parallel computing — a technique that spreads computational tasks across many computers. Google and IBM hope to advocate the cause by offering the considerable gear necessary to universities remotely.

[Not directly FOSS related but still of interest. - Scott]

Full Story

  Nav
» Read more about: Story Type: News Story; Groups: IBM

» Get new stories by e-mail: nightly or as they are published

« Return to the newswire homepage

  Related stories
» More like this

Comments (4)

tuxchick ·
What's the point of clustering? I don't see the advantage- why not go for a nice stout mainframe? A brand-new IBM z9 runs about $100,000, which isn't cheap, but then you have a single super-duper well-built machine. Instead of a herd of lesser machines, which are more complex to administer, presents more points of failure, and eat up more space and more power.
jdixon ·
> ...why not go for a nice stout mainframe?

Probably inertia and familiarity. When Google first started, they wouldn't have been able to afford a mainframe, so they went with a small cluster. That's what they've used and what they know.
bigg ·
One reason (as I've seen firsthand) is to use the computers they've already got more efficiently. A secretary's computer sits idle all night and all weekend, in other words, most of the time. If you start looking around, you find that most computers spend more time collecting dust than doing work. To the researcher with heavy computing needs and a limited budget, that is gold.
dinotrac ·
>I don't see the advantage

Mainframe's are wonderful things, but they aren't super-computers. Mainframes have always been about powerful I/O capabilities, distributed across umpteen controllers -- sort of like cluster computing for I/O.

Cluster computers can, for some problems, deliver super-computer performance.

Otherwise, though, I agree that mainframes make a wonderful answer to many problems that people throw piles of little servers at.

Post a comment

A free account is needed to post. Write your comment here first: your text is kept in this browser while you sign up or log in, and is waiting for you when you come back.

Your message:
|||||
10
Markdown: **bold**, *italic*, [text](https://link), - lists, > quotes, $x^2$ for math. Formatting help

Sign up and post (it is free)  or  Log in and post