Red Hat snatches storage Gluster file system for $136m

What's inside Pandora's box, in fact

Commercial Linux distributor Red Hat is paying $136m to acquire Gluster, the privately held maker of the GlusterFS cluster file system that is used by some hot properties on the intertubes.

Gluster is a spin-off from California Digital Corp, a supercomputer maker that was tapped by Lawrence Livermore National Laboratory to build its Itanium-based "Thunder" parallel supercomputer in 2003, which at one point was the second fastest supercomputer in the world. The techies behind it, led by Anand Babu Periasamy, Gluster's CTO, founded Gluster in 2005 to build a better file system.

The name Gluster is the combination of GNU and cluster, and will be a source of many typos over the years. The company hopes to pitch its open source clustered file system against the open source Lustre file system, formerly owned by Sun Microsystems and more or less ignored by Oracle, and General Parallel File System, a closed source file system created by IBM for its supercomputers. The Gluster file system was created with the idea that having a centralized metadata server lowers performance and limits scalability (as has happened with Lustre, in fact) and should make use of the underlying file system on arrays and not store data in proprietary formats.

Gluster put out its prototype in early 2007 and the first release, Version 2, was announced in May 2009. The file system is used by about 100 commercial companies now, including streaming music service Pandora as well as Box.net, Deutche Bank, Samsung, BAE Systems, Barnes & Noble, and Autodesk. By acquiring Gluster, Red Hat not only gains access to those customers, but also brings all the brains behind Gluster and gives them paychecks to keep doing what they are doing.

"Our customers are looking for software-based storage solutions that manage their file-based data on-premise, in the cloud and bridging between the two," said Red Hat CTO Brian Stevens in a statement announcing the acquisition. "With unstructured data growth (such as log files, virtual machines, email, audio, video and documents), the 90s paradigm of forcing everything into expensive, single-system DBMS residing on an internal corporate SAN has become unwieldy and impractical."

Periasamy said it better, and with the kind of humour you expect from someone who has just become rich: "Gluster started off with a goal to be the Red Hat of storage. Now, we are the storage of Red Hat."

Making a hash of it

The thing that has Stevens excited about Gluster is that rather than having a single metadata server at the heart of the distributed file system, GlusterFS has come up with a no-metadata server model, based on an elastic hashing algorithm, that allows GlusterFS to scale across more than 500 x86 servers that implement a distributed NAS, holding petabytes of data.

Gluster can ride on top of ext3, ext4, XFS, and other file systems and exposes the file system as a global namespace that spans the storage server nodes as an NFS or CIFS mount point. It can feed up data to Linux, Unix, and Windows servers and has multitenant capabilities that make it suitable for use in private and public clouds. GlusterFS can also be deployed on Amazon's EC2 cloud with its Amazon Machine Image (AMI) format or inside of KVM virtual machines. Gluster is working on some "big data" enhancements for unstructured data, and Red Hat is interested in that as well.

The other thing that Red Hat is excited about with GlusterFS is that it has a community of developers with over 2,000 participants. It is buying control of a community that helps it make better products, like Fedora does with Linux and JBoss does with middleware.

Red Hat says that in addition to the $136m in cash, the company will be assuming unvested outstanding equity in Gluster and will be issuing equity retention incentives to certain employees. Red Hat says that Gluster will have no material impact on revenues this year, but that stock-based compensation and amortization costs will go up between $1m and $2m on the deal.

Red Hat plans to sell GlusterFS on a subscription basis as well as integrating it into product stacks. ®