Since its launch DCIM (Data Centre Infrastructure Management) software has been a controversial topic. Potential customers and many commentators have claimed that the software is overly complex, expensive, overhyped in terms of its promise and underwhelming in terms of its achievement.
Nonetheless, careful management of the assets of a data centre, including the infrastructure that supports its core operations, is essential. Tools that can quantify and simplify the necessary tasks will always be in demand. Despite the disillusionment felt by some organisations, a well thought-out DCIM deployment strategy can deliver significant benefits to data centre management.
Recent rollouts of DCIM solutions at organisations as disparate as Cardiff University’s High Performance Computing (HPC) data centre, and TV-service provider Sky support the point. DCIM can help an organisation maximise the efficiency and utility of its infrastructure, provide greater insights into capacity issues before they become a problem and improve the environmental impact of its data centre.
Adopting solutions as complex as DCIM has not been without difficulty. Those that study the practice of introducing innovative technologies, whether they are university academics or specialist consultancies, have models such as the S-curve of innovation or the Gartner Hype Cycle which seek to quantify how the pace of innovation moves forward at different rates depending on the maturity of the technology in question, the perceived need for its deployment in the first place and the proven benefits that accrue once sufficient experience has been gained to deal with the inevitable problems that arise.
DCIM is no different. A class of software tools emerged to fulfil a perceived need: that of more closely monitoring and managing the assets of data centre infrastructure such as power and cooling for which there had hitherto been inadequate or ad hoc solutions. Naturally, the data centre world jumped at the opportunity that was being promised and the market moved very quickly to invest in the new systems.
Inevitably, the early adopters discovered that DCIM was not an instant panacea, that deploying and managing such tools often required a great deal more work than had been foreseen, that some of the expected benefits were not realised and that the necessary re-engineering of both technical infrastructure and operational procedures proved to be more expensive than expected.
A natural consequence was a slow down in adoption of DCIM as the lessons were learned and expectations were recalibrated. Armed with this hard-earned experience, more successful deployments were realised and confidence in the ability of DCIM to deliver genuine innovation and productivity kick-started further adoption.
So what have we learned from the experience of early DCIM adopters?
Well for a start, goals have to be realistic and achievable. To that end, a realistic assessment of the costs incurred by a DCIM deployment is essential. DCIM can only manage what can first be measured. Often, infrastructure elements such as generators are not connected to IP networks. Additional instrumentation to gather all the data needed by a DCIM solution has to be budgeted for and installed. A rule of thumb that “Your budget will be twice what you thought it would be” is not entirely inaccurate.
We must also realise that more efficient management of data centres is essential. Today’s increasingly virtualised IT installations in which computing capability is delivered as a service via the Cloud is producing a highly competitive market for those services, one in which control of costs is a vital competitive advantage. Controlling energy consumption, arising from both IT equipment and the attendant cooling infrastructure yields benefits in terms of reputation—we all want to be “greener”; and cost savings—however low fossil-fuel prices may be temporarily, it is always a good idea to guard against price fluctuations in this most unpredictable and capricious commodity.
DCIM is rarely a quick plug and play deployment. It requires significant preparation and adjustment before benefits are realised. Nevertheless the benefits of greater visibility and control become apparent once the software is in place.
Another consideration is just what tools one needs to deploy and to what degree one can make use of management tools already in place. Inevitably, there will be an installed base of systems management tools in place. A “forklift upgrade” in which all existing software is removed and replaced by a complete suite of new management tools may sound like a good option but in practice, some integration with an installed base is inevitable.
DCIM deployments frequently require significant changes to managing a data centre before they can be exploited and it is inevitable that they will drive further process changes afterwards. Management of this change is a major challenge for data-centre operators. However, customers and manufacturers are becoming more experienced and better informed.
To take the examples of Cardiff University and Sky: the former was very much focussed on delivering a green computing environment and succeeded after deploying an Energy Efficiency DCIM module, along with associated improvements to its cooling infrastructure, in reducing its PUE rating from between 1.7 and 1.8 to a level between 1.2 and 1.3.
The Energy Efficiency Module was an optional add-on component to the University’s data centre’s existing DCIM suite of software. Working with data inputs from extensive instrumentation, the new module provided a comprehensive picture, on a centralised console, of power and cooling consumption throughout the data centre infrastructure.
It allowed personnel to get much deeper, more granular insights into energy usage, not just at overall site level but also at subsystem level and, critically, it did so in real time. For example, it allowed them to monitor the effects on energy consumption of changing fan speeds, or of CPU utilisation on a server rack, or of raising the temperature of the chilled water supply.
Sky was more concerned with reliability; maintaining uptime of its 1000 racks of servers was its primary goal, closely followed by the desire to manage capacity issues and efficiency more effectively with the overall intention of reducing cost. It made use of a DCIM package containing modules dealing specifically with Asset Management, Change Management, Capacity Planning, IT Optimisation and integration with existing Business Management Systems (BMS) to achieve these goals.
Naturally the aims of both organisations dovetail to a certain extent. More efficient power consumption at Cardiff leads to a reduction of costs and Sky’s better utilisation of its existing infrastructure helps it achieve its own environmental goals, thanks in part to its belief that “the most efficient data centre is the one you don’t have to build”.
Examples such as these, and others, show that DCIM is becoming an established and trusted part of data centre management across the full industry range from large service providers to smaller installations, and is delivering real benefits to all concerned. The lessons learned by the pioneering generation are helping deliver the benefits of overall data centre management to all.