It’s really exciting to be able to share a new network service capability, this one is distinct and complementary to our existing Update Connection. I’d like to introduce the Eco Connection – power management for your data center assets. (Go direct using the link above, or navigate by going to My Sun Connection, click the “Fresh Tools” tab, then click the Eco Connection link.)

Here’s a snapshot of the capability at work.

Power Consumption for a server

Temperature of the CPU for the same server

This capability capitalizes on work done in Sun Labs around the continuous system telemetry harness conceived by Kenny Gross and team. What we do is collect telemetry from sensors resident in computers, establish baselines, then compare the telemetry patterns to patterns we’ve collected that are known good and bad.

For instance, if CPU temperature is climbing but power consumption and workload stay constant (as in the above images,) we have a nascent failure mode to identify. We can model this situation by covering an air intake with something as small as an index card. It turns out that MTBF is highly influenced by factors like elevated temperature over time – even slight elevation of temperature.

Another great advantage to this capability is being able to aggregate and trend power consumption at an application service level and data center level.

This isn’t finished, we’ll continue to enhance it over the coming days, weeks, and months adding data for our new T2000 and x2100/4100/4200 servers, power conditioning equipment, and non-Sun equipment in the data center. I invite you to visit often to see how we’re progressing.

One thing we’re hoping you notice is that we’re rolling these things out and letting you get a feel for them early in the development lifecycle. Expect to see more and different capabilities show up in the very near future. Tell us what you think, good or bad, and we’ll factor it into our plans.