The red hot "Big Data" space with Hadoop, NoSQL, Memcache and others continues to attract new players like moths to the light (or is it flies to honey?). Another player in this space is Gluster. Gluster is a Silicon Valley based, open source, software only storage technology provider. Gluster is the developer of the GlusterFS, "an open source distributed file system distinguished by multiple architectural differences, including a modular, stackable design and a unique, no-metadata server architecture". They are also the managers of the Gluster open source project.
The folks at Gluster believe that storage is going to go the way of computing. That is to the cloud, on commodity hardware and widely distributed. Their software only approach to storage is aimed at just that scenario. I had a chance to chat with John Kriesa, VP of Marketing and Tom Trainer, Dir of Product Marketing at Gluster about this vision. To them what we are seeing in terms of the Big Data market is only a taste of the future. Also key to the Gluster approach is to treat storage of data and objects. Their unique approach is to treat storage as objects. From the Gluster site, here is their vision of the future of storage:
Storage needs to catch up with other areas of the data center and accelerate, rather than inhibit, adaptation in this new reality.
Gluster believes that storage not only must support the new computing environment (virtualized, multi-tenant, scale-on demand) but will inevitably end up looking more like it (open source, scale out, commoditized, and paid-for on a utility basis).
So it would seem only natural then that Gluster would find a way to work with some of the technologies powering big data. Of course none is hotter right now than Hadoop. There are actually two parts to Hadoop. One is the so called MapR applications and the other is the HDFS file system. What Gluster has done is made it wasy to swap out the Gluster GlusterFS with the HDFS. Actually they don't even have to be mutually exclusive. You could have both file systems running side by side. So you can run your MapR app with storage in a GlusterFS or HDFS or both!
That is another part of the Gluster story. They are providing NAS like storage into these big data, Hadoop installs. Marrying this tried and true storage technology into these new age, distributed cloud apps give Gluster a unique place in this ecosystem.
The real question is will people swap out the default Hadoop system for Gluster? With a thriving community and many paid customers already using GlusterFS commercially I think there will certainly be a sub-set of folks who do.
In the meantime, news and requests keep pouring in on the Big Data front. Stay tuned for more news on this subject over the next few days!