Using software to save energy is not a radical idea.
So, how come it isn't core thinking either?
Software - automation really - brings the most powerful opportunity for optimization and energy savings. There are lots of storage software features and applications that help manage the placement of data, move it from one tier to another, prune copies, make incremental copies instead of full, and eliminate redundant data. Lots of choices and lots of opportunities to accomplish business ends and use less energy at the same time.
Yet, in my experience, discussion about energy savings center mostly on the efficiency of power supplies or bits of computer and storage systems. Of course these need to be made as efficient as possible. That's foundation work though. Even the most energy efficient of systems won't help you if not deployed and operated with all of their efficiency potential.
It's the software.
There are several vendors with point products to help here. EMC has has an edge with a whole portfolio that helps.
Think about it.
Snaps instead of clones.
Clones instead of Mirrors.
Virtual LUN migration.
Policy driven data migration for files or email.
Compression and deduplication.
And. And. And....
All of these, individually and together can help make the data center more efficient, use less hardware and use what you have more effectively.
In their August, 2007 report to Congress, the U.S. Environmental Protection Agency (EPA) estimates that data centers can save more than 20% on electricity through modest energy-efficiency initiatives. Add best practices and save 45%. Use state-of-the-art practices, and get up to 55% energy improvement.
This is using existing technology and best practices. And a lot of it involves automation made possible by software ranging from virtualization to data migration.
The point is that we know what to do, how do do it and with what.
How come we don't?
There are some legitimate obstacles. For example, potential operational disruption is always a concern when making changes and sometimes there are risks. But these can be managed and shouldn't be allowed to slow the pace of adoption. There are lots of examples where very savvy IT operations are using these approaches already and are already getting the benefits - including the long-term savings in energy use and operating expense.
Two things about this have become apparent to me. First, all of these software capabilities were typically created for some purpose other than energy conservation. So, awareness of their energy saving properties is low. The other is that the connection between physical devices using energy and how those devices are efficiently operated needs more emphasis as well as analysis.
I haven't taken a scientific survey but that's what the anecdotal evidence suggests.
This past week I spent time with a very experienced group of IT executives who are looking for ways to advance their competitive edge in technology. My assertion that software was the most important element in saving energy drew some skepticism.
So, we stepped through it. Design the most efficient platform. Make configurations scalable and flexible. Include the entire range of drive sizes and speeds. This is where the question is typically asked: "If everyone has access to the same drives, how does that make yours different?"
This is where it gets interesting.
In general, each manufacturer does have the same disk drive choices. How they are used (or even if all the choices are available) is the next question.
Take EMC's assertion of more efficient system design with a grain of salt? Fine. Set that aside and what's left?
Software. Automation. Wise operation.
Seeing the success of Apple in recent years, it is not hard to believe that they follow their own advice to "think differently". Steve Jobs, when asked if he was concerned about others trying to copy the IPhone, said he had no worry. They can copy the device he said, but Apple is five years ahead on software.
In the case of energy, efficiency and environmental impacts, much can be accomplished by thinking differently about what we already know.
Software is key to that. Existing software. No new technology required. In the storage realm, this is where EMC excels in breadth, depth and quality of the portfolio. (But always assume that the R&D machine is still grinding at full throttle.)
There are many examples. Deduplication and content management were discussed in an earlier post. Server virtualization with VMware is another obvious play. And, just as with server virtualization, learning to leverage capabilities of other applications allows users to dramatically reduce energy usage. File virtualization software is a great example that can provide huge energy savings.
File virtualization software optimizes file storage simply by separating the physical location of the file from its logical location in the file system organization. Since different files have different utilization rates, the file virtualization software allocates files to the appropriate level of tiered storage according to user-defined rules that monitor how often the data is updated or accessed.
This allows files with higher utilization rates to be stored on high performance, low-capacity drives. Infrequently used files are stored on large capacity, low-speed drives that are much more power efficient. This perfect allocation of resources results in greater utilization—as well as better power efficiency—that is completely transparent to performance.
There are lots of other options to consider for storage, servers, and networks. Individually, none will solve the energy problem, although each can provide immediate solutions for specific problems. Used together, systematically, you will discover very effective energy solutions that you really already knew but hadn't thought about in just that way.