David Appelbaum is vice president of marketing at Sentilla Corporation, and has worked in software marketing roles at Borland, Oracle, Autonomy, Salesforce.com, BigFix, and Act-On.
In Part 1 of this article on data center Key Performance Indicators (KPIs), I compared managing a data center to piloting a jumbo jet. Although an experienced pilot can fly a small plane by sight alone in good weather, the same pilot needs a flight plan, cockpit instrumentation, and air traffic control support in order to fly in foggy weather. Data centers today are akin to that foggy day, with new technologies and applications adding layers of abstraction and complexity to data center management.
In Part 1, I also described results from a 2012 survey of 5,000 data center professionals highlighting the metrics they use to provide visibility into data center infrastructure to support intelligent decision-making. Relevant metrics show not only what’s happening in your data center today, but also how it relates to other parts of the data center and to your costs. The following commentary references the survey findings.
Good Data is Hard to Find
If you are not getting the visibility and metrics to make informed decisions about your data center utilization and capacity, you are not alone. The 2012 survey asked respondents about KPIs for cost and capacity.
Many metrics that respondents indicated were most important were among the most difficult to get from existing tools.
This is evidence of a serious visibility gap – data center management tools do not make it easy to find important information around power capacity and costs, as well as overall operations and maintenance costs.
Puzzled about Power? Join the Crowd.
If nothing else, both Super Bowl 2013 and the widespread outages caused by Hurricane Sandy and the subsequent reliance on fuel-powered generators caused many to rethink their relationship with power. For one Manhattan-based data center, people formed a bucket brigade hauling diesel fuel up 17 flights of stairs to keep the data center’s backup generator running. Suddenly, excessive power consumption has a real, human cost.
The survey included questions about two important power-related metrics:
- Power density, expressed in watts/square foot
- Power Usage Efficiency (PUE) – a ratio of the power used in the data center to the power that is used for computing (versus heating, cooling and other overhead power consumption)
Of all the metrics-related questions in the survey, the two power-related metrics had the highest number of “Don’t Know” responses.
More than a third of respondents didn’t know their PUE.
This lack of knowledge might signal an organizational gap that is hindering visibility into necessary metrics.
Finding the Metrics Through Management Tools
The first wave of Data Center Infrastructure Management (DCIM) solutions only show basic facilities and IT metrics, such as PUE, energy consumption and temperature fluctuation. More recently, a new wave of DCIM solutions are emerging to fill the gap – by aggregating and correlating information from the other categories of data center management solutions using a “manager of managers” approach.
While relatively few respondents were currently using DCIM tools, a full 38% were planning or considering implementing DCIM.
The new generation of solutions will deliver not only better insights through global visibility, but also the ability to perform proactive project planning and evolve to KPI-driven data center management.
Flying with Instrumentation: Data Center Management Maturity
Having better visibility is the essential first step in moving towards a more mature approach to infrastructure management. With visibility in place, you can start proactively planning based on predictions and forecasts, and eventually move towards KPI-driven data center management.
If you manage your data center as if you were piloting in the fog, implementing a unified data center performance management solution can help you move up the maturity cycle.
- Start by monitoring and analyzing capacity and utilization of all of the data center assets (including virtual, physical and facility layers). This insight will help you cut costs, reclaim under-used capacity and avoid over-provisioning.
- With a unified data center performance management solution in place, move to a proactive planning approach: use predictive analysis and what-if scenarios to make smart decisions. This approach helps you optimize your IT spending and availability while alerting you to potential constraints before they cause availability or performance problems.
- With a firm understanding of business needs and data center performance, adopt a KPI-driven data center approach, enforcing service levels and understanding the true cost of services. Data centers in this stage of maturity are best positioned to align resource spending with business needs.
You Don't Have to Fly Blindly
Just as you don't want to fly that plane without instrumentation, you don't want to plan your data center capacity based on incomplete data and best guesses. But, getting these metrics means moving beyond the isolated enterprise IT management, virtualization management and facilities monitoring tools that you may have in place today.
By deploying a unified Data Center management solution, you can get the KPIs that put context around your planning decisions – integrating operating cost and power metrics with IT assets, and supporting predictive and what-if analysis. With this insight, you can reduce the cost of over-provisioning while mitigating the risks of unanticipated capacity problems.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.