Subscribe to RSS Add to Technorati Faves Digg This Page Send to Stumble Upon Bookmark on Delicious

Tips & Tricks from IT Data Center management

posted by Jeff Moriarty on March 06, 2007

In this post I’m trying something new (for me at least). Based on some of the ideas and input I’ve had for topics, I’m now wandering around Intel conducting some interviews. The intent is to showcase some of the other people, ideas, and programs inside IT beyond my immediate cubicle walls. My first victim in Project IT Blog Stalker was John Musilli, a Data Center Manager in IT.

John and I have been speakers together on the Intel Premier IT Professional engagements in several cities, and he definitely knows his space. John has had oversight over some of the largest data centers at Intel, and can probably calculate in his head the power consumption and air flow necessary to deploy a server farm at the top of Mount Everest with sheep as the only energy source.

The first thing comment John had on data center management was that the industry method isn’t always the best and cost effective solution. John said one of the biggest problems he sees in the data center arena is that everyone wants to see what everybody else is doing, instead of focusing on the core problems they have to solve. For example, in most data centers you see cables run from server to patch panel, patch panel to patch panel, then finally patch panel to switch. You can save a lot of expense by running from the server right to the switch with a single cable. John said a lot of the “industry standards” and recommendations come from the vendors. There’s nothing technically wrong with the standards, but they tend to favor the vendor’s components… like buying lots of patch panels.

John also emphasized that data centers do not have to be pretty. There’s often a push to make them sleek and good looking, but really you just need to meet the operational needs of the servers. If you go with a more “industrial look” you can save in a number of areas including the way you set up your racking, power and network distributions.

John warned that both of these areas can create some political challenges. Often the people who are setting data center standards are pulling from what they see others do, so advocating things like non-standard cabling methods and appearance can ruffle feathers. However they are great areas to save costs and improve the efficiency of your data center up to 50% cost savings.

The three points John said he would really want to make to anyone in the data center arena are: 1. Focus on your performance needs, not the industry standards. Don’t be afraid to question. 2. Blade servers are your friends. You will get much greater performance for the power and thermal footprint. 3. Your room may not be hot, just ill managed. If you think your center is running hot, evaluate your capacity, and then look to at server placement and optimizing existing cooling and air flows.

John is going to be watching this blog to reply to questions or comments, so let them fly. If this topic is interesting we are happy to do a follow-up or drill down in a specific area.

Comments (2)
del.icio.us StumbleUpon Digg It
tagged: , , , , , ,

Comments

Mar 06  |  James Smith said:

If I recall Intel has a hardware & software standard for servers, it used to be called P100. I think an explanation of P100 and the whole philosophy behind it including “copy exactly” as it applies to the IT environment would be useful for readers ..

Mar 06  |  Jeff Moriarty said:

James, good call and probably a topic worthy of its own post, but here’s the short version:

Intel has a general policy of “copy exactly” when it comes to our FABs. The idea is that if we get a high performing operation working, don’t reinvent it. FABs are hideously precise operations so take what you know already works.

The Intel IT version of this (on a much smaller scale) is a series of technical certifications necessary for deployment of an application or system into our network. It validates a program/system, its configuration, and its operation. “P100” is the name of the certification for general deployment.

In our data centers the idea is that all servers deployed should adhere to one of the approved P100 configurations. It helps reduce complexity.

I’ll dig into this more in a separate post, but let me know if you have any specific questions.

Post Your Comment




Comment Policy: We welcome your comments, however all comments are moderated. Offensive, off-topic or fraudulent comments will be deleted and not displayed. By submitting a comment to an Intel Blog, you agree to our legal information and privacy policy terms, including having your name displayed with your comment and that you are 13 years old or older. Your name and personal information will not be used for any other purpose, and your e-mail address will not be published.

Disclaimer: Opinions expressed here and in any corresponding comments are the personal opinions of the original authors, and do not necessarily reflect the views of Intel. All Intel names and trademarks are the property of Intel Corporation or its subsidiaries in the United States and other countries. Other names and brands may be claimed as the property of others.