San Francisco - Linux clusters are exploding in popularity for computer-intensive tasks, but until now the lack of a journaling file system has held back storage applications. Minneapolis-based Sistina Software's Global File System (GFS) may be the missing link. Its cluster nodes physically share storage over fiber channel or shared SCSI, and while each node thinks the file system is local, file access is synchronized across the whole cluster. In effect, GFS can pool storage onto cheap, efficient machines.
Sistina's VP of business development Dave Sass explained that GFS was an academic project at the University of Minnesota until the beginning of 2001, when the company was formed. Sistina, which now employs 44 people, has raised $5m - $3.5 of which was from local venture fund St Paul VC. Sistina executives are off to California to start the series-B funding round.
Matt O'Keefe, Sistina's CEO and founder, released GFS as open source software under the General Public License (GPL), meaning it's now very widely deployed. Sass estimated that there have been 9,000 downloads of GFS and as many as 14,000 of a related technology, the Cluster Logical Volume Manager (CLVM). "Without much marketing, our software has been adopted as the standard in Linux," he claimed.
Now Sistina's partners - and, more to the point, potential investors - would like the business model to change, as pure GPL plays aren't the irresistible opportunities they used to be. Besides, Sass added, the GPL can be: "frustrating to customers who want to make a little bit of money off their software."
That's why GFS will soon be available under the Sistina Public License, a derivative of the Aladdin Public License used for Ghostscript. As Sass explained it, download for non-commercial use is free. Redistributors and outsourcers - essentially, anyone making money off GFS - are required to pay Sistina a license fee.
Sass identified three different potential customer pools. First are resellers, exemplified by a recent deal with San Diego's Western Scientific. "The problem they were finding is that customers would want to use GFS but weren't sure what hardware to buy," Sass explained. The partners will work together on pre-testing cluster configurations for technical computing, Web serving, media streaming, video production and so on.
OEMs form the second logical customer group. "In fact we're having discussions with companies as large as Compaq about doing similar things," said Sass.
And third is the embedded space. There are obvious advantages to using a clustered file system with network-attached storage appliances.
"Essentially what GFS does is that rather than having islands of storage, our file system can be accessed for read-and-write from every server in the cluster. That means you can dynamically add servers in a granular way," Sass explained. "It works across heterogeneous platforms very nicely, though the advantage of Linux is that it's a single platform, from the embedded controller right up to the mainframe. We've done some work with IBM on their zSeries mainframe to have multiple instances of Linux with a single file system on the back end." It would be hard to think of a more cost-efficient system for xSPs.
Sistina is consciously imitating Veritas, which grew its business by distributing its file system with Sun's Solaris. But Sass added that, "At this point in time we don't view Veritas as competition." The real rivals are IBM, which has its own proprietary clustering file system, and EMC.
"EMC faces tremendous pressure on the margins," Sass pointed out. "It probably has the most elegant management tools on the market ? Veritas products don't even come close. That's the positive. The downfall is that it's proprietary - customers have to pay a large premium. Until now, enterprise customers have been willing to pay for serviceability and scalability, but when there are a lot of other alternatives out there at lower cost, that's likely to change."
EMC excels in the xSP market, and xSPs are coming under immense cost pressures themselves. "Any new technology that comes along that looks like it has the potential to commoditize storage, they're going to look at it," said Sass confidently. "Almost every large company we go talk to is experimenting with Linux. We're a little bit ahead of the market, with a clustered file system and with Linux. With IBM's investment and the new IDC figures, though, Linux is starting to look like the unified Unix that never came."
the451 (www.the451.com) is an analyst firm that provides timely, detailed and independent analysis of news in technology, communications and media. To evaluate the service click here.