Hardware Refactoring

Refactoring is always a good thing… Cleaning some code, make it faster, shorter, better. The same thing needs to be done with the hardware. That’s what Chris, Urban and I did last Thursday with our cluster in Oerlikon. Besides it looks much better now there are some other advantages.

We moved away some old servers and replaced them by (less!) new ones, such as the new poweredge 1950. We could reduce the latency of map.search.ch by 2 and we save maintenance, power and storage costs as well.

But there is one small inconvenience as well: heat. As the same power does use less and less power, the heat comes also on less space. We had some problems at the beginning, when the servers where one on top of each other (IO transfers on the harddisks started to get very slow on some machines). So we had to put some space between each server to guarantee a better cooling. Now everything works well!

cluster

Technorati Tags: , ,

3 Comments

  1. Silvan says:

    Niiiice. Andere freuen sich über Bilder von vollbusigen Mädels. Bei mir sind’s Bilder von Racks voll von potenter Hardware, welche mein Herz höher schlagen lassen…

    Cheers!

  2. Toni says:

    Those Racks look VERY familiar, looks like we are hosting in the same place.

    Now that we know your rack numbers, we can pay them a visit with our "<a href="http://www.thinkgeek.com/geektoys/science/770f/" rel="nofollow">Strong Magnets</a>" ;-)

  3. Gregor J. Rothfuss says:

    silvan: i hope you are aware of the phrase \’nice rack\’ :)

    denis: reduce latency by 2? factor of 2, or what? that is quite impressive.

    [ Yes, by factor 2, compared to our old machines... ]