* New data center requires new metrics
The face of the data center is changing, that much is certain. We see continuing and rapid data-center consolidation across many industries, higher-density technologies such as blade servers putting strains on heating and power systems, and efforts to automate many data center functions to reduce operational overhead. All of these trends create a changing landscape, which makes data centers harder to “read.”
For data center managers and facilities managers to get a better idea of the performance and “success” of their data center initiatives, we need to consider new metrics. Nemertes Research is examining some potential metrics for the next-generation data center:
* Data center density, as CPU cycles over square footage. This represents a measure of data center density, and therefore facilities utilization. Data centers with more rack density and higher space utilization score better. While useful in itself, it is also a very important factor in other metrics.
* Automation, as human operators over data center density. This metric provides a measure of data center automation. Fewer humans managing the same density of CPU-feet (see above) represent a more automated data center.
* Storage utilization, as percent of storage used. This is a tricky metric, because “storage used” is hard to define. Do you count storage that is allocated to an application but not yet used? Do you count high-speed “live” storage for apps and back-up/disaster-recovery storage on cheaper media as equivalent? What about unnecessary duplicates of data or empty records in a database? OK, so it’s not easy. But it can still be useful. If you define an accounting method for storage that is “right” for you and use it consistently, this metric allows you to track change over time, which can be very informative in times of rapid consolidation.
* Storage density, as storage utilization over square footage. Similar to the data center density, this metric reveals if you are making the best use of the facilities you have.
* Storage automation, as human operators over storage density. As with data center automation above, this tracks how efficiently you manage “utilized” storage for every square foot of facilities.
Whether or not these metrics are the best approach, it is clear that some traditional metrics may provide a misleading impression. For example, if you simply count humans per square foot, as some data center managers do, this makes huge and empty data centers look very efficient and packed blade centers inefficient by comparison. As the data center landscape changes dramatically, it is important to adjust the performance metrics to track these changes accurately. That way, you can really demonstrate the performance and automation gains you have made in your consolidation marathons without being thwarted by irrelevant metrics. Write in to tell us if these metrics would work for you or if you have better ones in use already. We’d love to hear your feedback.
