I've been talking to storage people in IT for about 20 years now. Sure, the new technology is always interesting -- but what storage folks are always interested in is how does this change things? How do things get done differently as a result?
And -- always unspoken but omnipresent -- how will this affect me?
While no aspect of enterprise IT is static these days, the storage team sits at the intersection of some particularly powerful forces these days.
Change is coming -- what does it look like?
It's All About Information, Baby ...
IT is required to play a powerful enabling role in helping the business do more with information across the board. Unpack the IT stack, and you'll find all sorts of interesting disciplines: application, security, networks, compute, finance, etc.
But all those 1s and 0s have to land somewhere, and that's what storage is all about. What makes storage so intrinsically interesting to many of us is that's where the information lives.
More Data, Bigger Data
As we chart the megaforces driving change in the model, we have to start with data itself, and how it's being used.
Everyone has seen the charts that show the constant dramatic increase in the amount of data that's being produced and consumed.
Digging a bit deeper, the data objects we're producing (images, video, logfiles, etc.) are individually getting much bigger. The cardinality of a collection of objects that we find interesting and useful (repository, database, etc.) is growing exponentially as well.
Bigger data elements in far bigger collections -- growth upon growth.
It used to be that applications largely "owned" their data, but no longer. One of the key tenets of big data (or anything to do with analytics) is that a data set has far more value when liberated from its source, and aggregated with others.
So it's not enough to think in terms of applications and their data -- information has become an asset that is increasingly disassociated from the application that created it.
We used to talk about the about Metcalfe's law regarding networks: that the value of a network was proportional to the square of the number of participants.
We're probably going to need a similar law that speaks to the value of data sharing across organizational and temporal boundaries.
I'm guessing it's going to be proportional to the cube of the number of diverse data sources.
Software Defined Storage Is More Than Marketing
Yes, flash is changing everything, but the far bigger megatrend is software-defined storage: doing in software once we used to do only in dedicated hardware.
Many of my colleagues have a two-part model; I have a three-part model I find more useful.
First, there's the data plane -- the stuff that persists data and makes it retrievable. Fail to get those bits right, and the rest doesn't really matter ...
Second, we have to consider an ever-growing long list of data services you can apply to that persisted data (caching, tiering, replication, move to cloud, differing presentations, encryption, archiving, indexing, deduplication, etc ...) that really ought to be considered separately from the data plane -- although that might not have been the case in the past.
And, finally, there is the all-important control plane: provisioning, orchestration, monitoring, management, etc. Control planes are integrated into multiple potential control points depending on the chosen management model: perhaps a separate storage control plane, perhaps integrated with the virtualization/cloud environment, perhaps the application owner, perhaps a federated model, etc.
What the control plane does is separate from the specifics of how it's presented and consumed -- and there is no "right" answer in my book.
Here's the headline: all three have clearly started to move to a software-only model.
We have newer software-only data planes to consider (insert your favorites here) that are a distinct departure from traditional arrays.
We are starting to see important data services starting to be decoupled from the underlying data plane or physical array.
And we're already in a world where storage devices can be almost entirely under software control -- either natively through rich APIs, or through one of the newer approaches such as EMC's ViPR.
Now, let's be clear, all three are hardly moving in lockstep.
For example, we'll probably see data plane and data services bound together for a while (usually in the form of physical storage arrays), and we'll be certainly using our newer control planes against these same familiar arrays for quite a while. And, of course, many of the newer software-only storage stacks are a bit light in certain regards, but that's just where we are today.
But it's started already, and it won't stop anytime soon.
ITaaS Is Real
The final powerful force we have to acknowledge is the new operational model of IT itself, dubbed ITaaS or IT-as-a-service for convenience.
The big idea? Design agile services that are attractive to consume vs. convenient to produce. Give people precisely what they need when they need it, and not before or after.
That notion can be applied to IT itself, or recursively to any subdomain. . I was talking with a customer who offered the opinion that simply tacking "as a service" to familiar entities was getting a bit trite. I pushed back -- it needs to be a constant reminder that IT consumption models have fundamentally changed, and IT is often behind the pace.
Fail to recognize this fundamental shift, and I fear that you and your IT organization will eventually become irrelevant.
Putting It All Together From The Storage Admin's View
So, let's summarize?
One -- information-centric business models, consuming vastly more data in ever-larger collections, increasingly separated from the application that created it.
Two -- a clear move to a software-defined model: either considering storage as a standalone entity, or part of a broader SDDC (software-defined data center) perspective.
Three -- an inarguable mandate for storage to be delivered as an attractive service: wide choices in data planes, data services and orchestration control points with no single winner.
The way I'd describe it is that storage admins are becoming storage service managers.
What's The Difference?
One way of looking at it is that storage admins have historically been focused on the back end of the operation: configuring physical arrays, zoning LUNs, provisioning resources, trying to manually align consumers with providers, dealing with vendors (!), cranky users and the rest of it. Trust me, these people are frequently being eaten alive by the enormity of their day-to-day reality.
Certainly not outward-facing, and not exactly empowered to think in terms of attractive-to-consume services that are dynamically invoked as needed, with flexible control plane integration depending on the consumption model.
The mindset evolves to "this is the suite of services my business customers desire -- how can I provide them as efficiently as possible?".
Maybe the data plane is running on a physical array, the data service is running on a physical appliance, and the control plane is the Oracle DBA. Or, alternatively, maybe the data plane is built into the hypervisor and uses server resources, the data services respect VM boundaries, and the control plane is integrated into the VM administrator's view. That's VSAN, by the way.
Or any combination that makes sense.
Once fully considered, that's a huge departure from how storage is done in so many shops today.
Where Are We Now, And When Are We Going To Get There?
Many of the first round of enabling technologies are starting to hit the market, with many more in the pipeline. Wait a few years, and we'll be swimming in potential software-defined storage solutions (data plane and/or data services and/or control planes) up to our ears :)
The folks who are already running private clouds in some fashion will be the first movers in enterprise IT. They've already got the required consumption and operational model working, they already appreciate the power of SDDC concepts, and -- beneath that -- software-defined storage. And the vast majority of them are VMware customers today.
But the real forcing function -- the move ITaaS and shifting the fundamental IT model -- is still playing out. Not everyone has been sucked into this particular maelstrom just yet (I met a customer who was starting to circle it just the other day, blissfully unaware of what lay ahead), but for those that are through and on the other side, they know what they have to do going forward.
Many of us got into storage because we thought it would be interesting.
So far, no one has been disappointed yet :)
Like this post? Why not subscribe via email?