Software Defined Storage - reducing storage friction
Back in 2012, when we made our storage predictions for this year, we remarked on the relevance of the "software defined" tag to the orchestration of storage resources: "Software defined storage will have significant impact on cloud, appliances and SSDs/Flash storage implementations. Software that is policy-based and treats storage resources agnostically will increase in order for data centers to become more adaptive."
So, how is this playing out? The addition of a layer of non-proprietary software abstraction creates the opportunity to manage storage more flexibly, which is a worthy goal. Software-defined storage has the potential to spell the end of over-provisioning as storage can be allocated more accurately from a wider pool. It also frees organisations from the cost overheads associated with proprietary storage.
Most of all it enables storage to be managed in a way which fits with application needs rather than the other way around. The historical challenges associated with storage allocation cause friction in more ways than one - not only are applications constrained, but the dialogue between application managers and storage managers represents an unnecessary overhead.
This being said, the process of getting there isn't going to be completely straightforward for a number of reasons - not least that it isn't easy to provision traditional, hardwired, LUN-based storage dynamically. Getting the existing, 'legacy' storage environment into shape will take time.
Even once this is done (in whole or in part), a second challenge emerges - the practicalities of data movement. Network constraints mean that shifting large quantities of data around takes time. As storage needs grow to cloud scale the problem gets bigger in parallel - potential applications are hampered by data friction.
A further challenge comes as a direct consequence of implementing centralised storage management capability. Not least that security and availability need to be considered across the whole storage environment, rather than providing assurances around individual storage devices.
Over the coming years, we expect to see storage become as easy to provision as processing and network resources, and therefore less and less of a bottleneck in the data centre. This opens the door to new types of application, for example, being able to quickly allocate all spare storage to big data analytics if it is not being used for anything else.
Software defined storage is less a case of offering infinite capacity - there will always be more to be done than resources available. However, by offering frictionless accessibility, organisations should be able to make much more of this vital resource.
Please share your thoughts below, what are the trends you are seeing in this area?