“No man should see how laws or sausages are made” – so goes the saying oft misattributed to Otto von Bismarck. This week it cropped up in the unlikely setting of the 451Group’s Hosting and Cloud Transformation Summit in London. Perhaps the first time either the Iron Chancellor or sausages have been associated with data centres. So I thought I would take it further and suggest that no man should see how power is monitored in the data centre. Why? Because power and infrastructure monitoring should be automated, taken care of by software, meters and sensors.
Relatively few data centres have implemented any automated energy monitoring solutions so far. But energy bills represent the biggest operational cost for data centres. There is general acceptance that something can and must be done about reducing energy consumption or, at least, slowing up its growth. So why the relatively low uptake levels of solutions?
I looked for clues in a session entitled “DCIM – The Elephant in the Datacenter”, Andy Lawrence described DCIM as one of the most hyped areas of the IT industry. However, he added that
If you want to get the best out of cloud, you need DCIM.
Speaking to fellow attendees, it is clear that the term DCIM means different things to different people. In practice, eating that particular elephant means a staged process and that means starting with automated metering and monitoring.
Much like fashion shows where the models wear outfits which seem unattainable or irrelevant to the average person, data centre conferences can be great places to hear about solutions that are at best aspirational. Hidden from the view of the audience are the safety pins and sellotape holding the fashion models’ outfits together, giving the illusion of the perfect line as they strut their stuff. So it is with data centres conferences. One operator described how its DCIM implementation integrates all the elements of managing their data centre estate – only to add that they still create spreadsheets to “share” information manually between elements (asset database and monitoring) of the DCIM system. Another illusion shattered!
A more mundane but very important issue discussed was lack of budget for data centre software. For example, allocating CapEx for metered power strips or branch circuit monitoring but not making provision for the monitoring system to leverage the investment in all that intelligent hardware. Why? A key reason cited is organisational silo management: IT and Facilities not collaborating and keeping budgets and objectives separate.
Others mentioned difficulties in defining RoI metrics; products look too complex to implement without a lot of (expensive) help. And “DCIM itself does not save energy” – may be but it helps you understand and then optimise operations.
Which brings us back to sausages! Put down the clipboard and the portable ammeter – how the power monitoring is done should be invisible and automated. Your focus needs to be on the outputs and the decision-making that follows from your new-found understanding of where the energy is being consumed.
Doing nothing is not an option; but neither is trying to do everything at once. Start with automated monitoring: either leveraging the intelligent metering you have already invested in or by implementing some (it can be done economically and non-intrusively). Once you have real-time visibility of live data, delivered as actionable information through dashboards and alerts, you are on the road to understanding. That is the critical step toward optimising data centre operations.
Through tuning and optimisation you make energy savings. You will see the effects of any changes you make to your operating environment. You will have direct feedback into your RoI metrics. And you will be able to achieve payback within a year. What’s not to like? What’s stopping you tucking into that elephant sausage now?