Despite everything happening around us, news is that we’re now up to 82% of capacity across Europe. That capacity combines energy, cooling and space. From personal experience I know it is tight, but because it is tight it made me go back to basics, and actually the problem is around process.
The amount of kit in a data centre at its simplest relates to processing and storage. The one thing that most businesses don’t do effectively is manage their data from a business perspective – in this respect IT has to take the lead and educate the business functions of the cost of data. I once worked out the cost of data as closely as was possible, it’s an inexact science but taking the cost of hardware, data centre cost, administration, security etc etc came up with around 50p per megabyte per year. That shocked people, and it was meant to!
People like data. They like the “power” it brings and often this brings a very personal behaviour with it. So it’s kept personally. The problem with this is that it spreads like a virus across the enterprise. Take for example a marketing department who prepare a new corporate presentation, for all the right reasons. It has wonderful colours, and great photographs promoting the business, and because its slick it’s a fair size file. Now that marketing department wants to share that with people and emails it to the top 500 people in the organisation. That increases the data by 500 fold. Now I know there are technical things that can be done to improve that – but how many organisations are structuring that data to minimise the space taken?
The other problem is “just in case”. This is usually a sign that a business doesn’t have a structured data retention policy that defines what is kept and why.
Also important is how that data is retained. Should data be in “near” storage or “far” storage? Near storage usually means on line spinning disks, far storage usually means off line storage media that occasionally may be needed, and likely can be stored away from the powered data centre.
So that’s some thoughts on data, but processing power is another big issue. A question I raise is; how many of you have actually looked at the percentage of time your server’s CPUs are running at full throttle? I’m guessing few and I’m also guessing the answer in the main is a low figure. The challenge is to look at how the processing is done and reduce the number of servers in use, because I guarantee you can run with less.
So there you have it – examine the data you think you should be storing and examine how you get the processing done with less kit and I think you’ll be surprised by how much you can reduce not just the space but also the cost.
Check out this videocast on data duplication and how to remove it.