Login:
Password:

IBM Opens GPFS File System to Linux

By Ed Oswald, BetaNews

December 14, 2005, 9:41 AM

IBM on Tuesday is expected to announce a new software strategy that will allow supercomputing customers to leverage the power of the General Parallel File System (GPFS) across a mixed-vendor computer cluster for the first time.

First developed by Big Blue in 2001, GPFS is the commercial name for the Tiger Shark file system developed by the company's Almaden research laboratory. Orginally built for work on large-scale multimedia projects, it was later extended to parallel computing applications.

GPFS supports files of several tens of terabytes and has run at input/output rates of several gigabytes per second, according to IBM whitepapers on the subject.

The company's new software strategy is expected to benefit computing applications that require large files, such as engineering design, digital media, data mining, financial analysis, seismic data processing and scientific research. The file system spans multiple servers and these large files can be executed from each of them.

Scientists have used GPFS to create and manage files that were hundreds of terabytes in size, the computer maker said.

The first vendor to license GPFS from IBM will be Linux Networx, which would use the file system to offer its customers a way to manage their data pools in supercomputing environments.

"As the popularity of Linux-based computing clustering grows, so does the need for simplified and highly performant file management software that is able to function across any hardware platform," IBM Deep Computing vice president David Turek said.

IBM said that it has continued to invest in the file system, making it faster and more reliable. Big Blue also has launched initiatives to ensure support on non-IBM hardware and will offer the GPFS source code to eligible clients.

IBM believes that its customers will embrace GPFS for supercomputing applications. "Customers will welcome the ability to use this superior file system across their data centers, in ways more flexible than ever before," Turek said.

Add a Comment (2 Comments)

BetaNews reserves the right to remove any comment at any time for any reason. Please keep your responses appropriate and on topic. Foul language and personal attacks will not be tolerated.

Name (required):

E-mail (required):

Enter Your Comment:

By horsecharles

posted Dec 14, 2005 - 2:53 PM

This project(which really does not impact us mainstream users too directly) is but one more example of ibm's commendable support of & allowing porting of competing linux to their latest technologies-- rather than limiting access to their own os'es only.
In particular the project which has the most direct & largest potential for us is the porting of Linux to the cell processor-- if this chip can somehow gather enough industry acknowledgment to make the jump from nextgen game console to the desktop, look out-- it is so many generations more advanced than the current top of the desktop 64bit crowd it's scary.
And this is where Apple has misstepped again(blinded & confused by both drm & sudden sales spike of a gadget that can overnight fall out of favor)-- instead of dropping ibm, it should've stuck w/ them & made the leap to this processor.
To those of you who may be wondering why the rant & big deal re this cell processor: are you debating whether to buy one type of processor because it's dual core, or its single core but multi-thread competitor? How would you instead like a combo multi-thread and multi-core(8 for now just in its first iteration!)
The problem is no one's making/planning to make desktops with this processor, so there needs to be a general outcry & demand for it, so that we're(mainstream users, not just a market niche) not stuck years & decades with inferior technology-- such as rudimentary pc's instead of workstation power(x86 platform instead of risc for more than a decade), & for several decades before that typewriters when pc's could've been available to us instead...

Folks: we as consumers need to demand & support the very latest developments, even tough it may not seem at the moment like we need or could put these inventions to good use--otherwise we may never get them. Quite often the company that invents something does not have the will, vision, or requisite skills to get a useful product to the masses-- witness how IBM & partners did not get the pc to us until decades later, and the risc processor never!
Sun has the technology to grid copious amounts of processors & ram together, while eliminating the need for a bus altogether-- but are WE going to ever get that???
WE NEED to support companies with innovative developments AND demand to get their products: ibm, sun-- whoever makes terabyte flash drives & holographic dvd's...otherwise we'll get stuck with comparatively low-capacity & slow computers, hard drives, compact discs....

MASSES: rise up & revolt-- refuse to be spoon-fed so much pablum & un-forthright marketing drivel!!!

Score: 0

By ArabianNight

posted Dec 15, 2005 - 4:25 PM

ummm...I stopped reading @ "This project"...

Score: 0