Wednesday, 31 August 2011

Maintenance Time (originally posted on 10th August)


One of the things that I hate the most about building and managing Data Centre is the fact that all maintenance on critical infrastructure has to be carried out at ungodly hours. Ally this to the fact that Uninterruptable Power Supplies (UPS) have factory recommended component swap-outs, whether or not the component, usually capacitors are broken or not.

It is an ass covering exercise all round. If they state five years and you don’t follow the recommendations they may last ten years plus, however if one or more of them go pop (usually with catastrophic consequences) then you get the old “Well we did warn you ” routine.

Last nights exercise included replacing the DC Capacitor banks on 2 x 200kVA machines. As they are A and B devices I could only do one at a time and it takes 90 minutes each. Throw in another 45 minutes per machine to carry out routine maintenance, and multiply that by 8 and as we had to start at 21:00 it made for a long night.

Good news was that we managed to finish the tasks one hour early, maintain the client load and put everything back where it came from (hopefully)

That was August

The year is passing us by very quickly and sadly, it will soon be Christmas. Summer holiday have come and gone and time to get all the Data Centre projects under way, and finish off the little nippy bits and pieces that are hanging about making productivity look bad.

My main focus at the moment is in adding to and upgrading our HVAC installation in our Imperial Way site. The existing units are in excess of ten years old and have performed admirably, sometimes over and above the call of duty in keeping the IT systems cool.

When they were designed, the criteria was for an external ambient temperature of 29degC. Sadly, in these days of Global Warming and Climate Change this figure now appears to be incredibly low. I am now looking at a design ambient of 35degC, which was unheard of a few years ago.

The issue is now how best to swap out the units without impacting the current cooling effort. Not only can we not afford to lose anything (again) but we also need to make space to install a further two new units.

This project will become a tagged update list as it proceeds