“It depends on what the meaning of the words ‘is’ is.” Famous words from a different context. But they kind of come to mind when you read about new technologies making their debut and the labels that accompany them. Software defined storage (SDS) vs software based storage (SBS) is a good example. Parsing the difference between the two terms could tie up a less technical user in knots (and I suspect some more techie types as well). When you compare SDS (or SBS) with software defined networking (SDN) (which is where the inspiration for SDS apparently originated), it appears to be found wanting. Just take a look at the lengths of their pages on Wikipedia and you cannot help but conclude that SDN as a concept has been developed better and is more advanced.
Storage is probably the last bastion in the IT infrastructure world to fall to the software wave and it appears to have held out much longer. Clearly no self-respecting storage admin will take risks with enterprise data, after all many a company has gone under due to data loss. However radical rip-and-replace change may be the long term solution needed to drive meaningful innovation. Currently SDS does not seem to fully deliver on this premise. Instead SDS often sounds more like a layer on top of current file or block-based storage systems, presumably making it easier to access them. Similarities with storage virtualization abound sometimes confusing the distinction between storage virtualization and SDS.
Preserving the status quo would certainly play to the advantage of the current leaders. After all lots of revenue is at stake. Yet a shakeup is probably what some users are hankering for, a loosening of the tight hold if you will that vendors such as EMC, VMware, and Dell have over the space.
Object based storage appears to hold the key, managing data as objects, with the use of commodity hardware to keep costs down. Making a difference is the REST API that could enable applications to talk directly to storage objects.
Lots of things are coming together: DevOps, the cloud, big unstructured data among others. DevOps and object storage look like a natural fit given that interaction happens via a REST API. Setting up a perfect storm for giving object storage a huge boost. Web-scale applications in the age of big data require a different storage paradigm. A paradigm that object storage seems to deliver on. However the “missing” piece of the puzzle is the open source storage stack.
What SDS appears to lack is an open source movement of the likes of the Open Networking Foundation. Where one has the option of implementing the entire storage stack with open source, software components that do not need to be purchased from commercial vendors (granted that many enterprise customers may not want to go down this path but this could open the door to a new batch of startups). No proprietary systems at all. OpenStack is definitely in play here but it is more centered on the cloud than on storage per se. Also there are some questions about how truly open it is.
Currently when you look at open source SDS most roads seem to lead to Red Hat: OpenStack (Cinder for block storage, Swift for object storage) where Red Hat plays an outsize role, Ceph, Gluster File System. Open source driven primarily by one vendor does not appear to meet the definition of what an open source ecosystem should look like. Moreover it has been reported that Red Hat is not averse to playing hardball with customers to use its own version of OpenStack by leveraging its dominance in the paid Linux market. So a different SDS movement seems to be called for, perhaps taking the cue from the networking world.
The elephant in the room appears to be the software used by Google, Facebook, and Amazon who have in all likelihood solved many of the challenges that SDS is setting out to address. Open sourcing some of these platforms like Facebook’s Haystack (of course this requires Facebook’s cooperation) might just be the trigger needed. That should launch the next generation of SDS focused startups that are unencumbered by the legacy approaches and technologies that characterize ones like EMC, HP, IBM, and Dell. All powered by a community and true open source, not the “closed” variety that is currently making the rounds.
Time for storage to break on through to the open side.