Three telltale signs that your data center is in distress
Time slips away. We see it in our everyday lives as our children grow older before our very eyes. We recognize it when our “new” car suddenly begins to feel obsolete.
Data centers experience much of the same phenomenon. Especially when this year's recommended infrastructure improvements are postponed until next year. That one year stretches out to three, and suddenly, the outside world has changed.
The data center is no longer flexible enough to react to the changes. The need to respond becomes acute and the company's technological competitive edge begins to wane.
According to 451 Research's “Voice of the Enterprise” survey, organizations are freeing up budgets and investing in modernizing neglected datacenter facilities.
Of those that are increasing spending, 37% are doing so to support datacenter retrofits or upgrade projects and 62% of organizations are consolidating their IT infrastructure.
The first step in rescuing an ailing data center is to recognize the signs of age. Below are some common examples of warning signs that point to a struggling data center:
Lack of space / power capacity – When data center physical space or power capacity begins to run out, two sets of problems emerge.
First, the business is constrained in its ability to grow at the desired pace. If a short-term market opportunity presents itself, whether or not the data center can adapt quickly is a critical success factor for cashing-in on that opportunity.
The second problem is that adding new servers can overload branch circuits and drive-up the data center temperatures. Both of these situations can lead to unanticipated downtime.
One low cost option for freeing-up data center space and/or power capacity is to reduce or consolidate IT load. This can occur either through virtualization or energy management (e.g., limiting how much electricity a server can consume).
Identification and archiving of unused servers and consolidation of servers that have CPU utilization below 10% are tactics that can reduce power consumption and, if remaining loads are physically consolidated, can save floor space.
Inefficient cooling – Over time, rack densities tend to increase. Eventually a tipping point is reached and hot spots emerge within the data center. Data center staff will need to determine if the existing cooling distribution can handle these more concentrated loads.
For a traditional, uncontained raised-floor data center, more than 50% of the cold air supplied from the cooling units will bypass back to these units directly as a result of leakage paths that exist.
These poor airflow management practices diminish the effectiveness of existing cooling systems.
Short term tactics such as adding blanking panels to racks, brush strips to raised floors, and row containment can increase the utilization of the cooling systems already in place.
If containment is applied, supply air will instead pass through IT equipment where it will absorb heat energy and transport it back to the cooling units.
The resulting higher exhaust air temperatures can increase the cooling capacity by 20% or more across existing cooling units.
Rising maintenance costs – When maintenance costs continue to climb at a steady rate, it could mean that antiquated systems are threatening the reliability of data center operations.
As a facility ages, the possibility of failures increases. Therefore, the development of an effective operations and maintenance program grows in importance.
Warning signs that components within the data center are close to end-of-life include vendors dropping support for installed systems, scarcity of spare parts, capacity and efficiency KPIs that don't meet current or future needs, and unserviceable parts are either failing or likely to fail.
Article by Martin Heller, Schneider Electric Data Center Blog