Customer and partner discussions seem to move in seasons. I'm sure there's some underlying cause-and-effect at work here, but sometimes it's a complete mystery to me.
Not to oversimplify, but I think that IT planners are starting to realize that whole topic of storage is going through a number of rather important transitions.
On the demand side, the evidence is clear: the majority of the organizations we support are seeing unprecedented demand to acquire, store and manage multiple rivers of information.
Put differently: here comes the flood.
On the supply side, there are significant and meaningful changes in how storage is built, how it's operated and how it's consumed.
Sure, there's lots of opportunities to do small things here and there to move along in the right direction: introduce a new technology here, fix a process there -- incrementalism vs. re-engineering.
But, at the same time, I'm finding more and more cases where it's probably time for our customers and partners to sit down and re-envision end-to-end how best to cope with the new world of information-intensive businesses.
First, keep in mind that I spend a lot of time with bigger (and more complex) environments. A lot of what I'm about to talk about here isn't applicable if your environment is more modest.
Second, there's a bit of negotiation that goes on in these sessions. There's always a list of topics that the audience is interested in as a starting point. But I also think there's a list of topics that the audience *should* be interested in, even if they say they're not :)
I also have to point out that this is how *I* personally tackle the topic as an individual here at EMC. Although you'll find a preponderance of EMC thinking here, I wouldn't consider it the final and official word on this topic :)
Finally, this is not happy-face marketing. These are serious discussions, often with not entirely pleasant implications for all involved. Better to get that out earlier rather than later.
A Quick Diagnostic
Not every IT organization is in this sort of pain. But there are quite a few where you can see it visibly hurts, and will only get worse unless a holistic approach is used.
Ask yourself these questions ...
What's going on in the business? Has there recently been a rapid surge in the amount of information that needs to be stored or retained? New businesses? New, large applications? New regulations that mean keeping more stuff around longer?
How are the people who interact with the storage team doing? Server team, virtualization team, app team, even exec IT management -- are communications and processes reasonably productive, or is there frustration in the air?
How's your story? Can you communicate your approach to storage using a handful of images? Services delivered, processes used, key technological elements? Or is it more like a bunch of stuff with a lot of ad-hoc processes around it?
If you aren't familiar with the frog-in-boiling-water analogy, it's useful here. Urban legend says that if you throw a live frog into hot water, it will immediately detect the situation and take corrective action. However, if you put the frog in cool water -- and slowly heat it -- the frog will never detect its peril, and eventually get cooked.
When there are highly visible crises of one sort or another, people can easily be galvanized into taking meaningful action. When it's a death of a thousand cuts, it's a lot more difficult to get the party started.
So, Where Do I Get Started In The Discussion?
A while back, I did a "mind map" of all the relevant topics that one might have in a "storage strategy" discussion -- diving down a bit, but not at an overly technical level.
The list of high-level topics alone went to six pages :(
Obviously, there's a lot that could be potentially be covered, but how do you organize the material so that you can have a meaningful dialog in, say, 60 or 90 minutes?
Keep in mind, everyone is coming from a different starting point. And, unfortunately, this stuff is way too familiar for me, so I can occasionally breeze over important points that I naively assume that everyone knows about.
So here's how I'm currently organizing the discussion:
Context setting, both in the industry and in IT.
Examples of industry context topics: a quick chat about expected information growth, shift in information types, new constraints on acquiring and using information, and so forth. Most technologists don't like to spend much time on these topics, but -- again -- a bit of big-picture is something I feel most technology companies don't spend enough time on.
Examples of IT context: pervasive virtualization, advent of ITaaS models (think "cloud"), new demands for speed and agility by the business, and so on. I want to make it clear that we see storage as only part of an end-to-end IT service delivery capability, and not an island :)
Key technological shifts in storage.
I don't want to argue the precise timing and status of these shifts, only that they're happening, and they're having an impact on how storage stuff gets built today and in the future.
As obvious examples, consider the shifts from tape to disk, and from disk to flash. Or the widespread use of industry-standard components vs. more proprietary hardware. The inherent appeal of scale-out storage architectures as information growth continues to outrace Moore's Law. Converged storage networks. Or storage value and innovation increasingly being expressed as software vs. hardware.
Nothing too controversial here -- all the usual stuff that most of us storage junkies sort of accept as the articles of faith.
I do throw in a few curve balls to make people think, though. One area I spend some time on is the importance of metadata in managing information, and how different forms of storage (block, file, object) can leverage metadata to differing degrees.
That's usually worth a bit of head-scratching, especially in content-rich environments.
The other area I spend a few moments on is "information logistics", basically getting the right information to the right place at the right time over nontrivial distances. As IT delivery models get more global, this is becoming far more interesting to more people.
Changing IT Management Models
If the technology is going to radically change, how we're organized to use it is going to have to change as well. One of my personal rants is that we -- as producers and consumers of IT technologies -- don't spend enough time around how we organize to consume the technology (and deliver services) vs. endlessly debating the merits of one technology vs. another.
My default reference model here is simple: storage services delivered in the context of other IT infrastructure services, with monitoring and control exposed upwards to other consumers of the "service". Sure, there are a variety of XaaS-ish models I've seen in larger IT organizations that work; the point here is to start thinking in that general direction.
Sometimes, I get an audience that is a bit defensive on this point. That's understandable. If I feel they're up for a little pushback, I ask them to explain the end-to-end process between a business user needing storage capacity, and them actually getting it.
Or how storage service delivery is continually measured for process improvement :) Again, this stuff isn't exactly rocket science, but it *is* a shift in perspective for many.
More importantly, your target organizational and management model will drive a very specific agenda in storage management tools: old school vs. new school thinking.
Data Protection, Replication and HA
Although there are interesting discussions that can be had at a topic-by-topic level (e.g. what's new in replication?) I'm finding more traction with putting all the "keep bad things from happening" topics on the table, and discussing them as a continuum -- mostly, since the lines are blurring fast.
For example, I can't see a clear line anymore between accelerated disk-based dedupe backup and continuous data protection (e.g. application journalling) -- you're just going for more frequent points-in-time at one level. Ditto for HA -- server and application failover is highly dependent on the data being available and usable.
Besides, the interesting parts of these discussions for me aren't the individual technologies, it's more about the integration points (VPLEX Geo, anyone?) and how management tools can turn the technologies into a "protection service catalog" that can be exposed to other entities in the organization.
And, if you think about it, that's sort of what most organizations want: a set of standardized (and integrated!) protection services that they can consume without worrying too much about the details.
Archiving And Tiering
I suppose this is an updated version of the old ILM (information lifecycle management) discussion, perhaps a bit more pragmatic in its current incarnation.
I try to make a couple of key points here to drive the thinking.
An example of mostly-metadata-free tiering approach would be EMC's FAST -- any knowledge we have about the data is externally supplied, and isn't intrinsic to the information itself.
There's a lot we can do -- even without metadata -- to boost performance and lower costs, but much more is possible if we've got metadata (associated directly with the information) telling us what to do.
An example of a metadata-rich archive might be email using something like SourceOne -- all sorts of rich policies can be generated simply because the data itself gives us some big clues on how it wants to be handled.
And, if you've got a big archiving/tiering challenge, maybe you ought to be thinking about generating useful metadata to make the automated policy management more tractable.
The second point I try to make is that archives have this sneaky way of assuming new roles over time. The archive that was built to reduce storage costs is now asked to do compliant retention. Or the archive that was built strictly for low-cost compliant retention becomes yet another information source for knowledge workers.
Once you have this ability to store non-trivial amounts of information all in one place, it's amazing what people might want to do with it :)
The final point I try to make is that the number and quality of external storage services is growing very quickly indeed. Just like in the physical world, long-term storage of digital assets may be something you'd like someone else to do on your behalf.
Not that there's a perfect solution out there for everyone, I just think it's something you should increasingly be open to.
Securing Data At Rest
It's rather difficult to isolate out a storage-specific security discussion -- simply because it's only a small part of the entire security ecosystem, but there are a few worthy topics that sometimes people are interested in.
One, of course, is the whole topic of encryption and key management, especially over the lifecycle of information -- backups, archives, etc.. The choices there are pretty clear, so really not much news there.
More interesting to many is the notion of securing multi-tenant environments -- not only protecting tenants from each other, but protecting them from the administrators as well. This latter topic goes into role-based credentials, audit logs, event monitoring and so on. Again, storage is just one part of the security ecosystem and has a role to play.
Finally, most security-aware environments need detailed GRC reporting: monitoring how the storage infrastructure is configured, administrative events, and so on. Lots to talk about there, but it gets very detailed very quickly.
Even with this generic framework, there are lots of interesting topics that don't really fit neatly into these standardized buckets. Yes, there's more. Much more.
One obvious example is advanced VMware integration: what's there, what's coming -- and how are people using it effectively? Perhaps more storage integration innovation is going on in that domain than any other. And there's more coming -- much more.
A less-obvious example might be all the similar integration work we're doing with other targeted environemnts: Microsoft and Hyper-V, Xen and KVM, Citrix, Oracle, SAP, mainframe, iSeries, etc. Yes, I know, we tend to talk a lot about VMware integration, but the actual integration story is far broader than most realize.
Project Lightning got a lot of press at EMC World, so I'm always putting that one on the table -- simply because it's very reflective of many of the underlying forces going on in the storage and broader IT infrastructure world.
If a lot of the audience's business involved shovelling content around the globe, I try and make time for the cloud object model as implemented by Atmos. It's not for everyone, but -- with every passing year -- more and more IT professionals are looking for a solution in this area, so it's worth a mention.
Along the same lines, I try to share our thinking around storage for big data applications: big data analytics (think Greenplum and Hadoop) as will as file-based workflows and applications (think Isilon). Both are very different ways of thinking about storage at scale, and usually outside the mainstream of enterprise IT applications.
Occasionally, there's some interest in VSAs -- virtual storage appliances -- basically, storage software stacks that emulate traditional hardware-based storage devices. While not exactly ready to solve the world's problems, they are very emblematic as to where the industry is heading: storage as virtualized software stacks that are pushed on your shared infrastructure pool.
And, yes, there's occasionally some interest in the whole consumer/SMB storage space, as embodied by what our friends at Iomega are doing. There's some cool stuff there, and it's usually very interesting to share just how far these folks have pushed consumer tech upwards into progressively more interesting market segments.
Service and Support Topics
There's still strong interest in topics like how we do interoperability testing in eLab (still a valuable service), or how we qualify hardware components (disk drives still fail, but less so), or how we interact with customer environments to provide customer support services -- especially in demanding environments.
There's also growing interest in professional and consulting services -- everything from help in formulating your own storage strategy, to designing/staffing/training the storage organization, to managed residencies and storage as a managed services. It's not a pure technology discussion, especially in real-world situations.
In each of these areas, there are interesting shifts and trends that are worthwhile to point to, especially in a planning context.
While every customer or partner situation is different, there are some generic food-for-thought recommendations we can make, e.g. start thinking about storage as a service vs. technology stacks, evaluating one or two of the interesting newer technologies to see if it makes sense in your environment, and so on.
I think it's worthwhile to give people at least a few high-level recommendations to take out of one of these sessions. We cover a lot of material, and it's nice to summarize at least to some degree.
Coming up with a specific, detailed and justifiable storage strategy -- well, that's usually a consulting engagement.
Obviously, there's a lot there, but -- that's the point -- there *is* a lot there, and it's all food for thought to some degree or another.
That being said, I'd be interested in any feedback you might have as to the approach here -- is it useful, could it be made better, are there important topics being missed?
One of the things that I feel fortunate about is that -- working for EMC -- there's a lot to talk about.
We don't have to force-fit one or two products into every situation we encounter. Sure, we end up talking about various EMC products and technologies, since those are the tangible examples of the thinking behind the strategic view.
As far as I'm concerned, the only reason we invest in these extended briefings is to help people get out of the day-to-day, take a moment to look ahead, and to start to plan for the future.
Because, when it comes to storage, the future is coming very fast at us indeed :)