Too much power is a bad thing

Not surprisingly, during the recent run-up in energy costs, power and cooling efficiency concerns once again became a hot topic with many of the customers I've been talking to.  And -- hats off to the server vendors who've realized that power efficiency matters.

Not surprisingly, during the recent run-up in energy costs, power and cooling efficiency concerns once again became a hot topic with many of the customers I've been talking to.  And -- hats off to the server vendors who've realized that power efficiency matters.

I think it works the same for everyone.  I live in New England, and my first gigantic power bill came across last January (right after the holiday season) and -- for the first time, really -- I started getting serious about turning off lights, powering down the computer, adjusting the heat down a few degrees.  When it's big enough to make a dent in your budget, you start paying attention.

In search of a magic bullet

But, interestingly enough, the storage discussion usually starts with "what is EMC doing about this for storage?".  There's a wishful hope for an exotic technology to help make the problem go away.  And if it doesn't take any extra work, it'd be even better.  Certainly this was the case in servers (more energy efficient processors, server virtualization, etc.) so why shouldn't this be the case in storage?

Yes, down the road, there are some potential opportunities for technology to do some of this for storage, but -- my apologies -- the real approach is to use today's technology a bit more effectively.  Do we need alternative fuel cars, or should we just a be a bit more judicious on how we use today's automobiles?  The answer is probably  "both".

Using what's there already

As an example, take today's enterprise-class 500GB drives.  All things being equal, a disk drive with twice the capacity is roughly twice as energy efficient as one with half the capacity.  Multiply this by hundreds (or thousands) of disk drives, and you've got a tantalizing potential to take a real bite out of energy costs.   

You'd think that -- given the explosion in information and the need to cut energy costs -- customers would be racing to the new technology.  But that's not the case, at least so far.

So why aren't people moving more aggressively to the bigger drives? I don't hear any solid answers.  Maybe they just haven't thought about it that way.  Or maybe they're not figuring energy costs savings into their ROI calculations.  Maybe they've got concerns about performance, or maybe reliability.  All that aside, I don't know about you, but if someone offered me a car similar to what I drive today that got twice as much mileage to the gallon, I'd certainly take a very hard look at it.

Maybe it's like public transit -- you know it's there, but you just haven't gotten around to figuring out how to use it yet.

Cleaning out the junk

Another candidate is stale and/or duplicate data.  As an example, we've got tools now that can spot duplicate email attachments and user files, delete the redundant copies, and give everyone a transparent pointer to the original. 

What will the savings be?  It's easy enough to find out these days -- simply load the tools and get back an estimate.  Sometimes people have been able to find 20-30% of their storage is dedicated to stuff that can be eliminated, or at least moved to a more cost-effective tier.  That can be a lot of drives that don't need to spin any more.

We're learning that file systems are a particular offender.  For years, there have been good products from EMC and others that can look at your file system, decide if the data hasn't been touched in a while, and roll it to tape leaving a pointer in the user's file system.  And an idle tape cartridge is kust about state-of-the-art when it comes to energy efficiency.

One of the recommendations the power company offered me on my bill was to clean out all the junk from the flues, venting, etc. to make sure that everything was working efficiently.  Maybe we should do the same thing with our file systems!

Once again, the starting point is understand your information and its value.  If you have information that's redundant, stale or never accessed, there's a great opportunity to store less, or at least store cheaper.

Don't count technology out ... just yet

Can technology make a dent in the future?  One area that is still being investigated is drive spin-down, e.g. powering the drive when no one is using it.  But most applications won't wait patiently for a drive to spin up when it's in that state -- the only exception being use cases like disk libraries where the backup application already understands it might have to wait a bit.  And I think more work is needed to really understand what continually power-cycling a drive does to it in terms of long-term reliability, but it's a start.

Another angle is data de-duplication -- spotting duplicated chunks of information and only storing the differences.  But there's no free lunch -- all the algorithms I'm aware of use a lot of memory and processing power to find and remove duplicate chunks, so that has to be factored back in.  And the underlying reliability of the storage environment has to be pretty robust in a data deduplication environment, as anyone who's gone deep into incremental backups probably knows.

Things aren't going to change that much

I think it's a pretty safe bet that energy concerns will continue to rise over time.  And I think that IT vendors will be much more aware of the issue as they look to their future products and technologies. 

I'm not counting on my power bill trending down over time, are you?  Just doesn't seem to be in the cards ...

But, on the other hand, nothing speaks as loud as customers voting with their purchases, so -- in many regards -- the acceleration of the industry is largely up to you: the consumer of these technologies.

Comments

[bob]

stupid