Energy Star for Software? by Zen Kishimoto

Green IT means two different things. One is to make IT itself greener, and the other is to make other things greener. Let’s focus on the former. As everyone knows, IT consists of hardware-such as servers, storage equipment, and network gear-and software. Up until now, energy efficiency has been focused on hardware but not on software.

I tried to cover this issue in the Green Software Unconference last August. See here and here.

In those posts, I listed several factors to consider in making software energy efficient:

• Parallel programming—hard to program, nondeterministic (such as race condition)
• Design/architecture—modularized structure vs. big monolithic chunk (e.g., SMTP server)
• Interpretive vs. compiled—web languages vs. C, C++, and Java
• Performance vs. ease of use—heavy on the ease of use, and the heavy lifting left to hardware
• Optimization—of compilers and other utilities
• Selection of algorithms

Although my session drew some audience, interest in software energy efficiency was limited. Why is this the case? First and foremost, most software engineers want to make their programs function correctly according to the specification. And then the programs will be optimized to run faster. Although optimized software tends to consume fewer computing resources and is likely to be greener, there is no guarantee that it is energy efficient.

What is energy efficiency for software? A more energy efficient server hardware box would produce the same results (like MIPS, million instructions per second) with less wattage. In other words, with the same amount of wattage, it produces more. MIPS is a standard metric for any server hardware box. If we use the same idea, then an energy efficient software piece would produce more results with the same wattage. In other words, an energy efficient software piece requires less wattage to produce the same results. The problem here is how we define the results. Software has no universal metric like MIPS for hardware.

	- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
		Energy Impact of Increased Server Inlet Temperature
	- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Two pieces of software could be compared for their energy efficiency if:

• They are run on the same hardware platform (same CPU, memory, and so forth).
• They produce exactly the same results.

There may be other conditions, but I cannot think of any at this point. This subject has been on my mind. Recently, I attended FountainBlue’s Clean Energy Entrepreneurs Forum on the topic “Leveraging Software for Clean Green Solutions” March 1 (5:30–7:30 p.m. at Applied Materials in Santa Clara.)

The focus of this meeting was on how to make other things greener with software. Software energy efficiency was too early to discuss because it is far more important to make other things greener at this stage. I will keep this topic in mind and try to make a point from time to time.

Tags: Energy Efficiency, Green Software

This entry was posted on Tuesday, March 2nd, 2010 at 11:02 AM and is filed under Community Manager, Green Technology. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.