Knowledge Is Power – And Power Management Needs Data

IT managers are more keen than ever to save money by saving power. But they can’t do it without first measuring what they’re actually using.

That’s clearly a passion for Ben Grimes, CTO of Avocent, a company best known for its KVM switches and Landesk client management software.We met him in London and got his view on why IT managers have to go back to first principles and meassure what they are doing.

We all know that saving power is an issue. But how global is it?
I’d arguethat it’s more important in Europe than the US, because there are parts of the US where power is cheap. where it was put in to power the textile industry which has now moved to the Far East. Here in Europe, it’s a big issue across the board.

IT managers need to save power. We all know that. Why can’t they just buy lower-power servers and hope for the best?
With the economy the way it is, CIOs want to get more out of what they have. They are questioning: can I maintain what I have, or do I have to build a new data centre?

But the problem is, they don’t have good infrastructure to make a good judgement. Most of them are using back-of-a-napkin calculations to decide whether they are out of power, or out of room. It’s not based on real factual data about what is being consumed in the server room.

When people calculate power consumption of their server room, it’s often based on the faceplate information from the equipment. I used to work at IBM, and I can tell you that we calculated the faceplate specifications by running the equipment at 100 percent utilisation, measuring it and putting in some sort of buffer.

Faceplate information is based on 100 percent utilisation, and in the past servers have been running at 17 to 20 percent utilisation. Now, virtualisation has increased that, but we are nowhere near 100 percent.
If you calculate your power consumption based on the faceplates of your equipment, you may build out more, when you still have viable headroom in the power envelope today.

A server running at 20 percent utilisation doesn’t use only 20 percent of the power of a fully loaded one, of course.
No, but it’s still lower than 100 percent. It’s north of 60-70 percent. If it’s less utilised it is not doing writes to onboard storage and spinning drives, and other things that crank up power. It also varies between workload.

And it varies over time.
You get spikes during the day, or during the year. November is busy in a retail environment, and during the day you get spikes in the morning when people come in and check their emails. You have to take all that into consideration when you design a data centre.

The technologies we have give you real-time information.

Page: 1 2 3

Peter Judge

Peter Judge has been involved with tech B2B publishing in the UK for many years, working at Ziff-Davis, ZDNet, IDG and Reed. His main interests are networking security, mobility and cloud

Recent Posts

Craig Wright Sentenced For Contempt Of Court

Suspended prison sentence for Craig Wright for “flagrant breach” of court order, after his false…

3 days ago

El Salvador To Sell Or Discontinue Bitcoin Wallet, After IMF Deal

Cash-strapped south American country agrees to sell or discontinue its national Bitcoin wallet after signing…

3 days ago

UK’s ICO Labels Google ‘Irresponsible’ For Tracking Change

Google's change will allow advertisers to track customers' digital “fingerprints”, but UK data protection watchdog…

3 days ago

EU Publishes iOS Interoperability Plans

European Commission publishes preliminary instructions to Apple on how to open up iOS to rivals,…

4 days ago

Momeni Convicted In Bob Lee Murder

San Francisco jury finds Nima Momeni guilty of second-degree murder of Cash App founder Bob…

4 days ago