Taking Cost Out Of The Data Center

With the recent shifts in the economy, there seems to be an entirely new interest in "cost out".

Sure, improving efficiency and taking capex / opex out of the IT budget has always been a popular topic, but it seems that it's become even *more* popular recently.

And there seems to be two major threads: a strategic thread about re-architecting IT (saving that one for another post!), and a more tactical flavor around "what can we do today?".

I thought I'd share with you the most popular quick-win cost-out projects we're seeing these days.

Why Cost Out, And Why Now?

OK, so it's not the most aspirational topic in the IT universe, but it's darn important these days.  I think one of the most motivational reasons for me is simple: money saved can often be re-invested in the business -- carving efficiencies out of today's IT environment helps fund tomorrow's new initiatives.

Besides, taking that cost-focused laser beam to as-is technologies and processes exposes all sorts of things that really ought to be fixed anyway -- and the cost-out rationale is as good as any to take on what you might have been putting off.

Same general story for most of the "green IT" discussion: a renewed focus on cost-cutting leads to efficiencies that many times directly impact carbon footprint.

And none of these suggestions are necessarily vendor or technology specific.  Sure, EMC does all of this stuff, and does it pretty well, but these tactical wins can be applied in most any situation.

So, without much fanfare, here's a quick rundown of the projects we're getting involved in on a day-to-day basis.

#1 -- Consolidate Your Storage

OK, so this one isn't entirely new, is it?  But it seems to be a treadmill: as IT gets their hands around one set of storage platforms to consolidate, more candidates seem to spring up like weeds.

When thinking about storage consolidation, most people tend to focus on the "higher utilization" benefits.  Sure, that's true, but I think there's far more to the story.

First, storage technology has gotten a lot cheaper to buy and especially to own.  Warranty and support costs are down, energy costs are way down, and so on.  Depending on how you account for older storage, you may be far better off running on new kit rather than keeping the older stuff alive.

Second, consolidated storage is easier to manage.  Unless you've got a well-defined storage management team, it's often the case that many IT groups touch the storage question: server people, test and dev people, database people and so on.  Consolidated storage implies consolidated management, and that frees people up to go do other things.

So, a question for you: how much storage has sprung up outside of your boundary?  When's the last time you ran a TCO model on your capacity, using current numbers? 

Depending on the situation, I'd guesstimate savings somewhere in the 15-40% range, depending on the degree of non-consolidation and the age of your equipement.

#2 -- Tier Your Storage

Another oldie but goodie.  I still meet many customers who haven't gone throught the excersize of creating a service catalog for storage, and exposing true costs back to the business.  We're not talking about a six month project here, more like a six-week quick win.

The usual result is a schema with 5-7 buckets, a defined SLA for each, and an implied cost per gigabyte that is easy to understand.

Regardless of whether you're envisioning charge back or not, the results are pretty consistent.  You put costs, growth rates and utilization factors in front of the business teams.  They look at it and go "wow!".  And an interesting discussion ensues about using less, using more efficiently, or both.

But the storage team has to start the discussion. 

Depending on the situation, I'd guesstimate the savings somewhere between the 20-30% range over a 24 month period.

#3 -- Virtualize Servers

OK, another blatantly obvious one.  And I know that many IT organizations are pursuing virtualization with abandon.  But -- at the same time -- I often see a certain hesitancy in the air that I feel is sometimes unjustified.

It's important to note that server reduction cost savings go beyond the servers themselves.  We're talking energy, cooling, floor space, network ports, etc.  And, if you can invest a bit in managing things differently, there's huge opex savings there as well.

Finally, if you're looking at disaster tolerant computing or business continuity, virtualizing your servers is probably the biggest cost-out technology I've seen in many years.

Some folks want to wait a bit longer and see what Microsoft or Xen can eventually deliver.  Fine, but the meter's running while you're taking that luxury.  Other folks are mistakenly concerned about I/O performance.  Categorically not an issue these days.  Or maybe there's concern that software vendor XYZ doesn't officially support virtual machines.

Whatever the barrier, it's always worth a closer look to see if you can accelerate the pace of change.  By the way, EMC has some practical experience in all of these areas.

I met one CIO at a mid-sized company who had one of his office walls decorated with power cords.  When his teams claimed they'd consolidated a certain number of servers, he'd say "bring me their power cords!" as proof of their elimination. 

I thought that was cool.

#4 -- Archive Aggressively

Here, we're talking file systems and email.  The technologies have been around for years, and there's lots of different ways to do this.  The basic idea is the same: find stuff that ain't so important, and put it in a place where it costs less to store and manage.

The larger your environment, the bigger the savings.  You'll hear crazy stats like 90% of all file data is never accessed after it's created.  Or that only 5% of your email really matters.  Intuitively, I can't argue with that line of analysis.

Actively archiving is a cost-out win on several levels.  Yes, you're moving stuff to cheaper storage -- that's always good.  Often, you're reducing redundant data, either through single instancing, or perhaps dedupe.

But you're also taking tons of data out of the backup/recovery/replication stream as well.  And managing archived data takes far less opex than actively changing data.

The costs and efforts associated with implementing these technologies has come way down in recent years, so if you haven't taken a look at the ROI associated with a quick-win active archiving project, maybe you should.

#5 -- Less Tape

This one's counterintuitive to most people, but the savings are there in many situations.  Sure, tape media can be inexpensive, but it's very expensive to handle.  Think big opex savings.

The TCO of disk-based tape replacement solutions have dropped dramatically in a very short period.  Sure, disk media is getting cheaper all the time, but now we have data deduplication (big win), and drive spin-down (another big win).  We're seeing situations where effective disk capacity for backup is actually getting cheaper than its tape equivalents.  And, if you go VTL, it's pretty much a drop-in solution.

And when you move to the operational aspects, people get freed up to do other, more useful work.

I'm meeting more companies that are coming up on end-of-lease (or end of useful life!) for their tape libraries, and don't necessarily want to sign up again for more of the same.  If you haven't run the numbers in the last 3-6 months (yes, it's changing that fast!), you should.

A quick note: I hadn't heard this before, but there are those out there who are spreading a bit of FUD stating that "tape is required for compliance reasons".  I bet I know where this might be coming from.

It sounded like nonsense to me, so I had some people go check it out, and the facts are pretty clear: there's nothing in any regulation (or any interpretation of regulation) that mandates a specific media type.  You just have to be able to reasonably prove that the data hasn't changed since it was written -- that's all. 

And that's something you'd have to do for both tape and disk -- as well as paper!

Understand True Costs To Assess True Savings

Our ability to measure true costs in the IT environment are severely limited in most cases.  The easiest numbers (e.g. what's on the purchase order) aren't necessarity the biggest numbers these days.  As energy costs and labor costs increase, I see people spending more time to get a handle on these harder-to-manage costs.

And Things Are Moving Fast

I also meet IT teams who may have looked at one or another of these topics in past years, and justifiably came up with a conclusion that the project didn't make sense at that time.  That's a valid decision.

But I think it's fair to point out that -- for each of these areas -- costs and effort have dropped dramatically.  None of these are rocket-science projects -- they're repeatable quick wins that have been deployed thousands and thousands of times.

So -- take a look at this list -- let me know your thoughts, and if there are any other quick wins I should be looking at.

Thanks!

Comments

[Mr. Tape]

Tape is good for off-site DR. Yes? Hard to take arrays/drives off-site.

[Chuck Hollis]

Sure, in some situations. I remember taking tapes home many, many years ago.

But that's a handling cost, isn't it? And tapes can go missing, right?

Network costs have come so far down, compression is so much more efficient, and disks can be used so much more efficiently these days -- so that's why people are more interested in alternatives to CTAM -- the Chevy Truck Access Method.

[mgbrit]

Letting your incumbent vendors know that you will be taking competitive bidding from the competition will also help you get the best product set at the keenest price.

Alternatives to this would be to lock in a minimum standard discount off list price with your incumbent vendors.

[Chuck Hollis]

Yes, of course ...

[Mark Fitzsimmons]

Hey Chuck,

I agree. This is a solid approach. We have posted to wikibon.com, with a high efficiency enterprise planning approach (HEEP). We have also read your blog response on Tech Target about chargebacks. We have rolled up into views information about the enterprise for both the IT and business partners that would greatly expedite these efforts. We are open to your comments. Please take a look at what we are doing.

Regards,
Mark