Interesting Reactions To EMC's 1Q08

Well, all things considered, I thought EMC's results were pretty good: strong growth across the board, new products doing well, share gains in many segments, etc.

But I guess a few people didn't see things the same way, e.g. "EMC Weathers Stormy Quarter, Denies Technology Threats". 

OK, so it makes for an attention-grabbing headline, but that's hardly the case.

But I thought some of the core questions -- and answers -- were worthy of a bit of further discussion.

Questions From The Analysts

One interest question from a financial analyst was along the lines of "gee, with all this thin provisioning and data deduplication, doesn't that mean that your customers will be buying less storage?"

The technically precise answer is "yes, generally speaking, virtual provisioning can help in some parts of some environments" and "yes, in some parts of some environments, data deduplication can help".

But I think certain people's expectations may be just a tad overblown ...

A Bit More Around Thin (or Virtual) Provisioning

The idea is simple: when an application needs capacity, provision a "virtual" view that might be considerably larger than what's actually being used.  As application footprints grow, more physical storage gets added as needed without the need to reprovision.

Now, beyond that, there's a ton of nuance about how to do this best without impacting performance, making sure virtualized capacity works well with things like management tools and replications, and so on. 

I'd be glad to get into that discussion, if you'd like ...

But the real issue is that provisioning additional physical storage capacity takes a certain amount of effort.   Because of this effort, there's a natural tendency is for administrators to provision more storage than is actually needed at a given point of time. 

How much more?  That depends.

Put differently, there's a tradeoff between management efficiency and storage efficiency.  Virtual provisioning shifts that tradeoff point for some people, not all.

I have encountered shops that take provisioning very seriously.  They've got great processes and great tools.  They tend to run at very high levels of utilization.  Thin provisioning is a nice option, but won't deliver a ton of benefit, because they haven't fallen into the all-too-common practice of "Have A Hunch, Provision A Bunch".

Other shops, well, maybe they're not so fortunate.  They haven't invested in the tools and processes, so they're in the habit of over-provisioning just to keep management effort down.  These people will benefit from thin provisioning, to be sure. 

They'd also benefit from better tools and processes, I'd humbly offer.

But, as Joe pointed out on the investor call, it's not a panacea in the long term, is it? 

Applications still require the same amount of physical storage at the end of the day.  Maybe you've picked up a few more points of utilization regarding better use of unused storage, but -- in the face of roughly 60% information growth every year -- it's a very small dent in the overall problem.

Other things are needed: tools that categorize information to help understand what's important, and what's not.  Better information governance policies that help IT keep what's needed, and ditch the rest.  And much, much more.

But, sadly, none of these are going to stem the tide of information growth.  Slow it a bit, maybe.  But the writing's on the wall -- just about everyone's going to need more capacity each and every year.

And, finally, virtual (or thin) provisioning is a feature of a product, not a product itself.  And, if you're a vendor (or an analyst!) trying to make a big deal out of this one capability, I'd strongly encourage you to reflect a bit around your overall approach to the market.

And Then There's Data Deduplication

I hate to restate old arguments, but I feel the need.  Yes, there's a fair amount of redundant data in most environments.  Various forms of data deduplication can help get it out, leading to storage -- and sometimes network -- efficiencies. 

Certain forms of backup are great candidates; certain kinds of file systems and repositories are other candidates, but we know that data dedupe won't help a lot with primary storage for a big SAP instance, or production DW environment, and so on.

EMC has gotten into the practice of studying specific environments before we make a recommendation on the value of dedupe, and -- not surprisingly -- there's a wide variation in the benefits.  Some environments, it's just plain obvious.  But there are plenty of environments where it just wouldn't be worth the trouble, all things considered.

Like virtual provisioning, data dedupe is a feature, not a product.  And, like virtual provisioning, it's a one-time hit. 

You get the excess out, but there's still plenty of new information being generated that has to be stored somewhere.

Netting It Out

The good thing is that the industry is looking at all sorts of different approaches to storing less information, storing it more efficiently, and so on.  All of this is good.

But, against a backdrop of the information tsunami, all of this will hardly make a dent in the big picture, as far as I can see. 

Actually, I see the opposite -- when it's cheaper to store more information, more information ends up being stored.  Watching what happened from 1GB to 3GB to 9GB to 18GB to 36GB to 73GB to 146GB to 300GB to 750GB to 1TB disks has shown me that -- despite better, cheaper or more efficient storage -- more money gets spent on storage, every year.

I'm betting that the same thing will happen with server virtualization -- when it's cheaper to run more server images, more server images will be run.

I believe the economics term for this is "positive elasticity".

Ever hear of someone who's figured out a way to store significanly less information year-over-year?  Or, perhaps run fewer server images than they did last year?

I haven't heard any of those stories -- yet.

Comments

[Conray]

It is all about cost.
When it is cheaper to just store what you have then it is to understand what to store you just keep storing more stuff.
It is like your closests at home, they just accumluate more stuff until they are full before we take the time to sort out what to keep or find bigger closets.
The real issue is that the time has come to understand what we have and what we want/need to keep.