Reliability and zero downtime -- or as close to it as you can get -- are now givens within the data centre. Instead, what's top of datacentre managers' minds according to recent surveys is achieving higher rates of utilisation of their existing assets, usually by means of several forms of consolidation including virtualisation.
So far so conventional. But is it really so simple? And if so, how do you get there?
According to Liebert's vice president of power engineering Peter Panfil, availability is no less important than it was. But alongside that priority sits the notion that the datacentre has to be ready for whatever comes next.
What exactly that future consists of remains unknown of course -- so flexibility is key, says Panfil. A crucial part of that flexibility is the ability to expand and reconfigure the hardware infrastructure as required, according to Panfil. Naturally, for power company Liebert, the key is to ensure that the power infrastructure is adaptable.
What that means, says Generac Power Systems' Mike Carr, is that datacentre managers need to get to grips with the engineering concepts involved in power generation and management and ensure that they have the skills on tap that allow them to manage the datacentre power systems to suit requirements.
What does that mean in practise? Carr, whose eponymous company makes AC generators, reckons enterprises should ensure that they install redundant power generation systems. This can mean, for instance, multiple generators installed in parallel. While this has meant expensive switchgear in the past, for Carr, his company's integration of such electronics into its smaller generators cuts the cost of redundancy. Similarly, APC's distribution panels include switching capabilities and the company's white papers stress the need for the existence of and visibility into standby generators.
It also means ensuring that the UPS and generation facilities are properly combined, so that staff are aware of how the systems are wired together. This ensures that, when the worst happens -- or even when a test takes place -- power to the servers is maintained.
In other words, what's happening now is that the discussion has moved on from simply providing power at all times to ensuring that power systems are configured in such a way that ensures consistency of approach. This has advantages both from the point of view of daily management and when it comes to re-configuring the data centre -- a process undertaken increasingly frequently.
And does it need saying that documentation of all configurations and changes are essential? This of course ensures that, as one IT manager experienced when a mail server went down, that you don't lose your entire Friday night figuring out which version of the hardware is installed and how it's set up.