Object storage and the systems that leverage its unique data storage approach have been the subject of much hype over the past few years. If you took past analyst predictions at face value, by now all unstructured data would be residing on an object based storage platform. For some reason, however, reality has not caught up with all the current hype. What then is the fate of object storage? Will it carve out a well defined niche or is it destined to be one of those technologies that barely makes it beyond a PowerPoint presentation?
Object Storage Backgrounder
As we discussed in a recent article, “What is Object Storage?”, object storage is a technology where data is stored in self-contained entities called “objects”. Think of an object as a file. Unlike traditional file systems, this storage method is not dependent on a hierarchical layout of directories and sub-directories. Objects are given unique ID numbers which are managed in a ‘flat’ index of the metadata (data about data) needed to store and retrieve a file. For details on what defines object storage, you can read the above article and the list of articles at the end of this entry.
The State of Object Storage
It is hard to tell if object storage is living up to its expectations. Our sense is that adoption is going slower than what the original object storage vendors had anticipated. First, nearly all the initial iterations of object storage required that an application be first modified in order to access the object store. While this modification was not overly complex, when implemented it allowed full exploitation of the object store’s advanced metadata capabilities..Nonetheless, most organizations didn’t have the time to undertake the modifications or they weren’t convinced that the potential gains object storage offered were worth the investment.
Cause for Concern
There is a justifiable cause for concern as various object storage vendors try to re-position themselves for success despite the realities of the market. First, we have seen a fair amount of change in leadership at more than a few of these vendors. As we discussed in an earlier entry, “What Does It Mean When A Startup Changes CEOs?”, leadership changes can result in revenue growth and other tangible measures of success, as was the case with Cleversafe’s appointment of John Morris as CEO. On the other hand, they can also be a warning sign of potential problems; such as we have witnessed from some other object storage vendors that have changed out their executive teams.
Another cause for concern is when an object storage vendor suddenly morphs into something else or no longer wants to be described as an object storage vendor. For example, we have seen more than a few of these companies offering iSCSI, CIFS or NFS overlays on to their object storage architecture and position themselves as tier 2 storage devices. While this is not necessarily a bad thing, it is a sign that the core message for the object storage technology is not getting out.
Pockets of Success
Despite these challenges, there have been pockets of success. Object storage companies, or at least divisions within existing storage companies, have a few things in common. First, in most cases they are offering a turnkey solution that includes hardware and the object storage software. The software-only model has seen limited adoption as far as we can tell. Interestingly, this is also true in more traditional storage. Integrated solutions continue to be a dominant way storage is purchased, even with all the hype around software defined storage. As we discussed in our article “Does ViPR Eliminate Vendor Lock In?”, the reality is a software-only approach has many of the same ‘lock in’ risks as an integrated solution without the benefit of a ‘single throat to choke’.
Second, most of these successes have come at the sweet spot for object storage; the hyperscale or web 2.0 data centers that need to store vast amounts of unstructured data. A few of the more successful object storage companies seem to be focused on these potential customers.
Finally, we are seeing a gradual but increasing success by companies that were doing object storage before object storage was “cool” – most notably, disk archiving vendors. The disk archiving space seems to be picking up its growth pace and the requirements for this use case fit well with the capabilities of object storage. Plus, the vendors in this space can easily bridge the development gap typically required to integrate to an object storage system.
What Needs to Change
While it may sound contradictory to the above comment on the focus on hyperscale data centers and the archive use case, ultimately object storage vendors need to “mainstream” the technology to make it more broadly appealing to the traditional data center. Data centers of all sizes now have an abundance of unstructured data that they want to keep for a variety of reasons. They will eventually face the same issues with traditional NAS that the hyperscale and large archive companies have had to deal with.
However, we don’t think this mainstreaming is going to be accomplished solely through building traditional file system gateways. While having a NAS or CIFS front end is a good start, direct application integration through file-sync-share products, archive solutions, collaboration solutions and direct application integration will provide a clean object storage interface that can be fully exploited.
This means that object storage vendors will need to get together and develop a ‘rising tide floats all boats’ philosophy to drive interest in their offerings with the mid-market data center. When storage managers are then convinced that object storage is the best way to store unstructured data, competing vendors can shift their efforts towards pointing out the differences between their solutions and other offerings on the market.
We don’t think the fate of object storage is in doubt. Success of these platforms is more of a “when” than an “if”. We believe widespread adoption will take place. Accelerating user acceptance, however, will require object storage providers to embrace a common theme. Perhaps something along the lines of: “Object Storage – The Best Place To Store Unstructured Data”. This effort would be similar to how the Active Archive Alliance helped saved tape. Additionally, this will require repeatedly emphasizing the specific weaknesses of traditional file systems, as well as promoting the development of a tier 2 (mid-range) data center set of solutions as was discussed in our article, “Unstructured Data Storage For The Mid-Market”.