According to data collected by the Uptime Institute (link here), up to 10% of the server workload in data centers are performing "duplicate" or "unnecessary" workloads. So, while folks are laser focused on optimizing PUEs, there could be substantial savings in simple rounding up and reducing workloads and servers assigned to projects that are no longer valid.

But how is an DC admin supposed to know what's "valid" or not? One of the potential interesting insights from a smart lighting system is the ability to help inform on space activity. While space activity alone will not dictate or prove server "validity" (or what I call freshness in the title of this blog.), it can be a useful proxy to go examine and investigate those servers more.

Here's an example: if rack 4 north, section 34 has not seen motion in over 4 months, those servers could be either working on projects that are entirely managed remotely, -or- working on an assignment that doesn't need to be visited frequently, -or- is a project that is stale, old, or the assigner has left/quit/moved on to other projects. In other words, motion can "inform" on freshness and validity. It's not a 100% telltale signal, but given the opportunity to use this flag as an invetigation point, could be very constrcutive in futher optimizing data center utilization.Of course, this type of insight more or less demands a high sensor count, otherwise, you won't get the necessary specifity to know which racks need investigation.

Interestingly, I could conjecture that this same heurisitic (no motion in an area for x amount of time), within a tight range of focus, could apply to warehousing, healthcare, life safety and security, retail applications, etc.

(Extra credit to anyone that can email me [jeremy at redwoodsystems ] the movie from which I used for the image in this blog entry.)