As I mentioned in a previous blog post, I've spent some time recently speaking with engineers and customers about power and cooling. I've heard a number of interesting power and cooling stories. Some of the stories are things that many people can relate to, since they may experience the same type of problem themselves. Needless to say, HP wants to help our customers who find themselves in these too common situations.
At one non-production data center a few years ago, there was a contractor working out on the street running by the facility. The good news was that the facility had redundant power feeds, but that wasn't an issue since the power was just fine. However, on this summer Friday afternoon, the contractor broke a water main, and water to the whole campus was turned off.
Since this was an older, non-production data center within a business campus, when the water was turned off it also impacted the data center cooling. With all the servers running in full power mode, needless to say, temperatures in the data center started to rise. Since no one really knew what was running in the data center, a too common occurrence, an e-mail went out to people on the campus to shut down any servers that they owned in the data center. Unfortunately, being a Friday during the summer, not as many people were in the office as there usually would be, and while some systems were shutdown gracefully, other systems had hard shutdowns before water and cooling was restored.
This story is specific to a place and time. However, I've heard enough similar stories from around the world that similar situations to know that this is something that needs to be considered for most data centers.
The HP Insight Control power management team had this and many other customer stories in mind as they were developing theData Center Power Control functionality. Essentially, it allows you to categorize systems based on their workload priority. If you have a power or cooling emergency, you can activate pre-defined rules that change the power states to save power or even gracefully shutdown the operation system and power off the server, even if people aren't physically in the data center. This allows you to ensure that critical systems can keep running while non-essential server are powered down.
Naturally, while this functionality was designed for power emergencies, it could be used for other purposes as well, such as shutting down systems over a long weekend to save power.
So, have you every had a power or cooling story that you'd like to share? Would power management functionality that was described here be useful for you?
Jacob Van Ewyk