What is the future of cooling capacity optimization in the data center?

This audio was created using Microsoft Azure Speech Services

On a recent trip to Schneider Electric’s European Technology Center, I asked Henrik Leerberg, Product Line Director for StruxureWare for Data Centers what he foresees for the future of capacity optimisation in data centers, particularly as far as cooling is concerned.

Capacity management and maximization has been one of the biggest drivers of the demand for DCIM tools for the last few years. It’s very, very difficult to do this manually, that is to attempt to understand manually how the connectivity between different devices is working especially with respect to how this interacts with the cooling performance within your data center.

Typically and historically, a lot of capacity management has been to do with recommendations for manual changes operators can make in their data centers. However with cooling, we’ve now reached a point where we can do this automatically using software and temperature data. So we can have a demand response system to use with the CRAC units and in turn with the entire air conditioning system within the data center.

In practical terms that means we’re automatically able to adjust CRAC units to ensure we’re supplying enough air to any area that’s getting hotter or cooler.

Where customers have less than a temperature sensor for every third rack, we’ll need to install more otherwise it’s simply doesn’t provide enough granularity for us to work with. With appropriate levels of information we can analyse a temperature map and use it to adjust temperature settings within the data center.

I mentioned to Henrik that I’m aware that a number of manufacturers have attempted to implement dynamic cooling before but not successfully, so I asked how this would work for customers

Henrik told me that he believes this feature could help save customers a great deal in energy costs as if you walk into a data center and find it cool, whilst they’re not doing anything necessarily wrong, they may well be spending more money on cooling that they could. Such a data center may be over commissioned or have the CRAC units running at full speed, which is very expensive for energy use and in carbon emissions. Henrik sees opportunities to go into facilities like this, make sure they’re instrumented correctly and then control temperatures automatically reducing emissions and operating costs.

The system is designed to have significant redundancy built in. This means that if it detect that a component in the system has errors or fails, we can switch to a bypass or fail safe mode. This could, for example, turn on all of the CRAIC units in the room and whilst it then may well be too cold or colder than is most efficient, it’ll make sure the servers aren’t interrupted by this issue and therefore minimizes the risk of any particular issue escalating or cascading into a bigger problem.

This new functionality is also to reflect that StruxureWare for Data Centers has a wide and growing range of features that have been specifically designed to ensure that customers who deploy the suite can harness operating cost savings as well as enhancing monitoring and management of their facilities. Additionally we have a range of technical information including whitepapers explaining how DCIM can help cut operating costs and how to avoid errors in either evaluating or implementing DCIM in your data center whether or not they choose Schneider Electric as their DCIM vendor.

Tags: , , , , , , ,