Comment "Software-defined storage" is something end users love, and the industry is going precisely in that direction. The only problem I have with it is that, as happens with other buzzwords, the term is being over-abused, and generalization leads to confusion.
I know that I'm not the first to raise this problem, but now that real software-defined storage solutions are appearing in the market, the terminology issue has become more problematic. As always, marketing is the first to be blamed, but finding someone who is NOT responsible for this situation can be quite difficult.
According to Wikipedia, in networking "software-defined" means:
Software-defined networking (SDN) is an approach to computer networking that allows network administrators to manage network services through abstraction of higher-level functionality. This is done by decoupling the system that makes decisions about where traffic is sent (the control plane) from the underlying systems that forward traffic to the selected destination (the data plane). The inventors and vendors of these systems claim that this simplifies networking.
In my humble opinion, the last sentence is the best part. To simplify, SDS means a separation between the control plane (where things are decided) and the data plane (where things happen).
The only two examples of true software-defined storage that I'm aware of are EMC ViPR and Primary Data (but feel free to leave a comment if you think differently). The first one has failed miserably because it just wasn't enough – I'd say rudimentary and incomplete – while the latter has all the potential to revolutionize the way large enterprises can do storage.
I've already talked about Primary Data in the past (comparing them to Nicira – now VMware NSX), and given that I've met them twice in the last two months I can confirm that, on paper, the solution is really great. Now that the product is available, it's up to them to demonstrate what they can really do. Looking at the various demos available and at the features they already have (and are promising), I think they deserve a lot of attention.
By the way, if you look at Primary Data's home page and collateral, they are the first to be uncomfortable using the term "software-defined." It isn't mentioned, and even though the Wikipedia definition fits very well with what it describes on its home page, Primary Data doesn't use it. I can understand why; it doesn't want to be associated with a meaningless word.
Server- and software-based storage
For the rest of the vendors playing in "software-defined storage," I have to say that they are doing an amazing job most of the time – but it's just a different job.
I like software installed on one or more servers to build storage infrastructures. It brings huge benefits to the end users (freedom, agility, lower costs, and so on) but there is no real separation between the data plane and the control plane here – everything still happens in the same place and "in-band."
To me, both VSA and hyper-converged players should be considered server-based storage. And it's only a matter of defining their real positioning, not about benefits they are bringing to end users. I'm not questioning the latter at all.
I'd also like to point out that almost every solution on the market is software-based today – even EMC VNX and NetApp FAS are software-based, aren't they? But it is also true that they are neither server-based (no off-the-shelf commodity servers are involved) nor software-defined.
Closing the circle
I know this rant won't change anything. It's disappointing, but I'm willing to live with it.
Server-based and software-defined storage are changing this industry and they are both well-appreciated by end users. They are simply two different things – putting them together is wrong and confusing.
Back to Primary Data; I've expressed my admiration for their ideas and vision more than once. At the same time, their vision is so disruptive for the enterprise space that it's hard to think they'll ever be able to succeed without some help.
As happened to Nicira a few years back, I think that companies like Cisco or VMware should consider acquiring them. Primary Data could be the perfect component to realize the end-to-end software-defined data center vision of these players. Especially for Cisco, which has no enterprise storage system at the moment, Primary Data could be the perfect component to become a primary storage player in the data center with a product that "rules them all." ®