It's got to be one of the most popular "grabber" lines in the industry, right?
And who doesn't want to save money on storage?
On the average, corporate capacity grows 60% per year. And, even if media prices fall 20-30%, it doesn't take a math major to realize that -- every year -- you're gonna be spending more money on the round, brown spinning stuff unless steps are taken.
So, here's what I've been telling people.
It's Not Really About Products
Yes, some vendors do a better job of offering physical media at attractive prices than other ones.
But I think there's no sustainable and substantial competitive advantage here -- we're all playing from the same bag of tricks -- at least, in the long term.
- All of us vendors have access to the same industry-standard parts bin: disks, memory, CPUs, power supplies, etc.
- Sooner or later, we all have the same tricks in our products: thin provisioning, big drives, etc.
- And, if you're willing to trade expertise and support for a rock-bottom price, that's your choice. But you're not going to get an ultra-cheap product and still get great interop, expertise and support. Those things cost money for everyone.
At best, maybe you're super sharp at playing the vendor game, and you can save maybe 5-10% off your storage costs.
But, even if you turn this into a full-time career (and people have tried), you're usually leaving much bigger savings on the table. You're stepping over dollars to pick up pennies.
And it has very little to do with products or vendors.
The Service Level Catalog
The single most powerful tool I have encountered in the last five years to help people save money is the boring old service level catalog.
This is nothing more than a simple grid that categorizes storage service levels (performance, availability), and assigns a user-visible cost to each GB stored per month or year.
3 categories probably isn't enough. 20 is probably too many. And it doesn't talk about vendors, or RAID, or anything else.
It's a service at a price.
But going through this simple exercise has proven to have wonderful effects.
First, many IT organizations really don't know what it costs to offer up a gigabyte of storage to the business. Kind of embarassing, but true. If you're smiling and agreeing as you read this, well, that's not a bad place to start.
Second, because IT doesn't know what it costs to offer up a GB, neither does the business. So it's hard to expect people to make rational tradeoff decisions when they don't know the costs involved.
And, if everything is free (or the same price), well, they'll usually take the best-looking alternative. Which inevitably leads to higher service levels (and costs) than can be justified.
You don't have to have a charge-back policy to make this work. You just need to expose the costs to the people who are storing the data so they at least can make some useful tradeoffs.
Third, once you've gone through this exercise, you usually find out that most of the information that needs to be stored and accessed is pretty low-service-level stuff. Almost active archives.
And, all of the sudden, there's a rationale to buy less of the speedy, reliable stuff and more of the slower, less-robust stuff.
Now, we're not recommending that data loss is acceptable. But there's a fair difference between someone who's decided on smaller drives and RAID 1, and someone who's decided on larger drives and RAID 5 or 6.
A noticable difference in cost, and a noticable difference in energy footprint.
This does not have to be a long, convoluted process. The templates and methodologies are pretty easy. EMC, as an example, has probably done this hundreds -- if not thousands -- of times.
On a personal note, we have a wine cellar at home. There are some very nice bottles in there, if you get my drift.
And the way that we keep those very nice bottles around is that we have lots more of the decent, inexpensive stuff racked up right next to it, so when we're fishing for a bottle to open, it's easy enough to reach for the cheap stuff, and not pull out a bottle we'd regret later ;-)
Same idea here.
As far as results, I've seen companies save anywhere from 10-40% of their annual storage expenditure by just performing this simple exercise, and letting natural processes take their course. It's that simple.
And it has absolutely nothing to do with products, or does it?
The Second Order Effect
There is a ton of new storage technology coming into the market. Big, fat drives. Thin provisioning. Data deduplication. Flash drives. And a whole bunch more.
Each of these is in reality a new candidate to deliver a particular class of service. Thin provisioning, as an example, has service level characteristics that many applications (and users!) would find, ummm, unacceptable.
Customers who've done the tiering exercise, and have a decent service level catalog, have it much easier.
When a new storage technology comes along, it's pretty easy to go to the service catalog, and say "hey, look at that, there's now a better way to our Bronze Level Service" using whatever has come along.
Compare that with someone who doesn't have this tool. They have to spend a lot of time figuring out what it is, what it means, where it goes, etc. And, of course, they run the risk of putting the new technology in the wrong place and suffering the consequences.
But there's an even bigger tool at hand, if you're up for it ...
ITIL For Storage
Sad to say, many IT organizations have been hit by the storage explosion bomb. Storage growth has been so rapid, they really haven't wrapped their arms around this from a strategic perspective.
Some organizations don't have a dedicated storage function. Or, what function they do have is mostly administrative -- simply servicing new requests, and keeping the boat afloat.
What they haven't done (yet) is step back and realized that (a) this stuff is growing like a weed, and (b) it needs the same sort of ITIL-like disciplines that other areas of IT are now enjoying.
EMC has built a useful set of transformational services to help customers do just this -- tackle storage as a strategic service to IT, and not an afterthought.
And all those great SRM and archiving tools really make a lot of sense when you've got an organizational function that knows how to exploit them.
Rather than go into the gory details, I'll just share with you a few of our counter-intuitive learnings in this process.
The Chain of (Storage) Lies
So, we saw this story over and over again, and it goes like this:
Business Guy needs a new application. Application Guy asks him/her -- how much capacity do you need? Now, Business Guy has no idea, of course, so he/she makes up a Really Big Number -- let's say, 200GB.
Now, of course, Application Guy doesn't trust the Business Guy, so he/she decides to double the number and pass it along to Database Guy, or Server Guy.
By the time it gets to Storage Guy (who doesn't trust any of them!) it's become a Really, Really Big Number -- maybe a terabyte or more.
And most of it sits there, overprovisioned and underutilized.
Is this a problem of technology, or people and process? I would argue the latter.
The Hoarding Phenomenon
Making something really hard to get makes people hoard. It's a consistent and predictable human behavior.
If there's strict controls on getting office supplies in your office, I wouldn't be surprised to find that everyone has their secret stash of pens, pencils, paper etc. squirreled away in their office.
Conversely, if there's always a ready source of office supplies in a central repository, you take what you need and leave the rest.
Storage is no different.
I have seen the hoarding behavior manifest itself in strange ways: people filling up capacity with junk so it looks used, or other gaming behavior.
The result is simple: if storage is easy to get, you'll use what you need, when you need it, and no more.
The result is that companies that keep 10-20% "swing" capacity around, and make it available without a lot of fuss, have far better utilization rates than folks who make it really, really hard to get.
No big surprise, when you think about it.
So, How Will You Save Money On Storage?
Will you pursue the cheapest provider, or the latest technology?
Or will you take a step back, and focus more on people, process and politics than just technology?
It's my belief that doing the latter will make you far, far better at doing the former.
Comments