LAS VEGAS. The world of storage is undergoing a transformation with new object-based storage technology.
As opposed to traditional file system block-based storage that uses file and directory hierarchies, object-based storage is a distributed system. While object-based storage has some inherent benefits, it also has some drawbacks, too. In a panel at the Interop conference in Las Vegas, object storage vendors explained the pros and cons.
“The use case for our object storage systems are for massive archives that are active,” Chris Gladwin, CEO of Cleversafe said. “Generally, what we find is that object is well-suited where you need petabyte scale and it’s well suited for video and image objects.”
Gladwin added that object is a great fit for deployments that have a need for limitless scale. Cleversafe has been in the market with its solutions since at least 2007 when it began to commercialize its own open source distributed storage network system.
Mark Goros, CEO of Caringo, told the Interop audience that his company was co-founded by the developer of one of the first object storage systems. That technology was sold to EMC and is now part of EMC’s Centera solution. Goros noted that Caringo is continuing the evolution of object based storage.
“Object storage is set up to do what file system storage can’t do,” Goros said. “It’s set up to store large amounts of unstructured data and that is most amount of content that needs to be stored today in the world.”
Bret Piatt, director of product management at Rackspace, said that his company’s Cloudfiles system is an object-based storage system. Cloudfiles is built using the open source OpenStack project, which was started by Rackspace and NASA in 2010. OpenStack has since grown to over 60 members including Cisco, Dell and Canonical.
Piatt noted that in the next decade there will be 50 times the amount of data that needs to be stored and it can’t all be put on block storage systems.
There are, however, a number of challenges and barriers to adoption for object-based storage. Goros noted that one of the biggest problems is inertia, as people are used to the block model.
“The actual world is changing but we still things in files and folder hierarchy,” Goros said. “But I believe a shift is happening as we speak and the world is moving to a flat hierarchy.”
Gladwin noted that there are things that people assume are limitations of storage that aren’t really true with object storage. He commented that file system storage systems are not designed for the idea of multiple geographic writes. In contrast, with object systems you can have simultaneous writes that enables things like Facebook where there is no co-ordination of the storage writes. Another key differentiator according to Gladwin is the ability to grow object storage systems.
Among the challenges that face object storage today is the fact that standards are still emerging. The vendors all noted that their systems use standard REST APIs and other HTTP type mechanisms to help create what have become defacto standards for object based storage systems
The security model for object-based storage is also different.
“With file systems you have directories and you can assign users and groups and everbody understands that construct, but that doesn’t map one to one with object storage today,” Piatt said. “That’s probably something we’re not doing very well right now that is hurting adoption.”