Imagine you were trying to build a great car engine.
Sure, there'd be a great discussion around different kinds of engine technologies, design considerations, fuels, etc.
But at some point, you'd want to understand a bit more about the car the engine is going in, and -- more importantly -- what the owner expects from the car.
I'm finding that the "storage strategy" discussion has many similarities.
The conversation can become very inward focused, and can sometimes lead to isolated approaches that don't exploit the context.
And I continue to see pretty big opportunities that are missed, simply because the lens isn't wide enough.
Why I Think This Is Important
I think I've mentioned that the best part of my job is that I get to spend a lot of time with customers.
And, somewhat frequently, a customer assembles a team to revisit the overall "storage strategy". And, we're lucky enough to meet with them.
Sure, there's a bunch to talk about within the well-defined confines of storage technology. Tiering. Archiving. IP storage. Virtualization. Better SRM. New approaches to backup, recovery and replication. Data reduction. Power and cooling. Lots and lots of storage products.
All worthwhile of discussion. And it's good stuff. And EMC probably has more to say than most on this entire topic.
But, more often than not, the really big storage levers are outside the confines of a crisp, politically acceptable storage strategy boundary. There's spill-over into servers, networks, classification, ITIL, business processes and so on.
Just like an engine is an integral part of what makes a car good, storage is an integral part of making IT good. And I think you can only get so far by considering it in isolation.
To make matters worse, many of the industry voices fuel this perspective. They tend to talk about what's in the box, rather than commenting on the best way to use the box by considering adjacencies.
So, let's talk about a few top-of-mind examples I've seen.
The Chain Of Lies
Here's how the problem starts ...
During a meeting, a business type gets asked by someone "how much storage capacity will you need?".
Now, of course, the business type has no idea what the answer might be, so they guess a really big number, maybe 200 GB.
The application person knows how hard it is to get more capacity, (and suspects that the business type is somewhat clueless) so he or she does the safe thing, and doubles the number to 400 GB.
The 400GB number gets relayed to the database person, who knows that this is only a guess, and it's really hard to get more storage, so they do the safe thing and double the number again, maybe to 800GB this time.
Database person relays the number to storage person, who knows how hard it is to get more capacity, so they bump the number up again, maybe this time to 1.5 TB. Or more.
Interesting to see how a harmless SWAG becomes a monster, simply because (a) there's no end to end view of the process, (b) there's no trust between the participants, and (c) it's really really painful when you guess wrong.
Of course, this never happens in your shop ...
Let's change gears a bit.
If I was planning to trek across the desert, I'd want to bring a lot of water with me, because it's really hard to get, and it's not fun if you run out. But, around my usual haunts, I don't carry 20 litres of water with me "just in case".
Imagine if additional storage capacity was really easy to get if you needed it. No need to inflate estimates. If you run out, more is available, and quickly. You wouldn't need to overprovision or thin provision "just in case".
I know this sounds crazy, but we've found more than a few customer cases where making it really easy to get additional capacity means you end up using (and wasting!) far less.
No need to overprovision. No need to hoard. No need to game the system.
Of course, this means that you've got some decent reporting tools on who's using what, what's getting full, and the ability to bring more online in days, rather than months. And all of that is very achievable with a variety of products on the market.
Back to our example of the "storage strategy team".
Imagine how silly they'd sound if they declared "we're going to save the company a bunch of money on storage by making it really easy to get more when someone needs it". Flies in the face of conventional wisdom, doesn't it?
But it's human nature at work. Think about it.
Archive The Junk
There seems to be a fundamental human genetic issue with throwing stuff away. If you've seen my basement, you'll know what I mean.
Information seems to be no exception. Making an irreversible decision to delete something so that it never, ever comes back is right up there with fear of heights, public speaking, and so on. Why don't we keep it, just in case?
I've seen storage strategy teams burn many cycles on trying to get people to delete stuff.
My practical reply is -- why don't you focus on archiving the junk as cheaply as possible? Don't fight the fight that goes against our genetic programming -- you may never succeed.
This brings up a discussion of information classification (files, emails, database records), movement to low-cost media, transparent access to users, and so on.
Oh, by the way, this sort of approach makes backup/recovery a whole bunch easier, remote replication much more cost-effective, gives you a leg up on compliance, retention, enterprise search, and so on ...
Lots of good capabilities from EMC and others in all these topics, but -- strictly speaking -- not within the well-defined bounds of a storage strategy.
More of an information management strategy.
So the storage strategy team moves on to something that's within bounds.
But that's where the big savings might be ...
Beat Up Your Vendors
This is nothing new. I saw this very much in play 20 years ago, and I don't think it will ever go out of fashion. And, all posturing aside, I think a healthy tension between customers and vendors is a good thing all around.
Keeps all of us on our toes!
But I don't think it leads to order-of-magnitude cost savings.
Look, we all use pretty much the same components to build our stuff. And we all have pretty similar cost of sales, and support costs, and so on.
Storage is a wickedly competitive marketplace. No one gets to overcharge anyone for any length of time.
Sure, a better negotiating strategy with your vendor can save a bit. But sometimes I get the feeling that people are stepping over hundred-dollar bills to pick up some loose change.
My favorite variant is the new, smaller storage vendor that claims order-of-magnitude cost savings with their TCO study.
Folks -- parts is parts. It's how you use the stuff that leads to the big savings.
Use 20-50% less storage due to better processes -- now that's real savings.
Or learn to classify the majority of your data and put 70%+ on the real cheap stuff -- now that's real savings.
But, unfortunately, the storage strategy team often doesn't look outside the narrow confines of their charter.
Too bad.
Should We Be Doing This Ourselves?
I don't see many customer-produced "cost of storage" charts that include manpower. Probably because it's hard to get to -- lots of people touch storage-related issues, and not all of them are on the storage team.
Going a bit farther, getting and keeping good storage people (and the good operational environment) is not an easy task. Those of you who are good at what you do, you know what I mean.
As an alternative, over the last few years, EMC has been pretty successful with an offering (managed services and/or residencies) that basically let you consume storage as a service, or (as with residencies) storage expertise as a service.
It's more popular than you might think. And the efficiency numbers of these environments stand in stark contrast to the run-of-the-mill IT shop.
How many storage strategy efforts even ask the question "is this something we should be doing ourselves?".
Might lead you to a different set of answers.
And A Few More Advanced Topics ...
Any time you look at any of the fundamental processes associated with IT management (discovery, provisioning, service delivery, chargeback, etc.) it's very hard to have a productive discussion just considering storage as a stand-alone entity. I believe the real efficiencies come when the tasks (and the underlying model) are integrated and not disparate.
And not too many storage strategies I've seen consider the storage outside of the data center -- branch offices, laptops, etc.
I think the most expensive (and risky!) terabyte you own is the one you don't manage.
Sorry For The Rant ...
We live in a connected world. It's getting harder to consider one thing without considering the things that connect to it in some way.
And, I think, in an effort to simplify and make progress, it's a basic human tendency to compartmentalize a problem, and analyze it in isolation.
Maybe that's a good way to keep your sanity.
But the more I work with customers on this specific topic, I keep coming back to the view that -- just maybe -- the best way to consider a storage strategy might be to focus on the car and its driver, rather than the engine itself.
Comments