By Jason dePreaux, Research Manager

I recently returned from AFCOM’s Data Center World, held in Orlando, FL. For an analyst that covers data center products like I do, shows like Data Center World are vital to keep abreast of the latest trends that are shaping the industry. Last year, I noted that efficiency was undoubtedly the biggest buzz word. I still feel that is the case, but I am realizing that efficiency is a very broad topic and it touches many different aspects of data center. One area in particular caught my attention at this year’s AFCOM show was data center monitoring and the emergence of DCIM products.

The basic idea with monitoring is that in order to reduce electricity use, you first must be able to measure it. With data centers, the chief metric is Power Usage Effectiveness (PUE). It compares the total electricity use of the data center facility with the electricity that ends up to the IT load. The closer the PUE is to 1.0, the more efficient the data center is. PUE has become something of a “how low can you go” competition, which is perhaps an oversimplification. Nonetheless, it is clear that metrics like PUE are helping to raise awareness on monitoring in data centers.  Besides saving on electricity costs, demand response rebates from utilities are popping up in certain areas like California, to provide additional incentive for data centers to invest in monitoring hardware and software.

Power monitoring can be done at many different points in the data center. I was somewhat surprised to learn at the conference that the most popular spot to calculate IT load currently is at the UPS. UPS tend to be upstream of the servers in a typical arrangement and perhaps not an ideal measuring point, but every data center has them, so it is a convenient start with little or no investment to get a handle on PUE. More detailed power information can be obtained further downstream at the branch circuit, rack PDU, or even individual server level. My feeling is that these monitoring points are becoming increasingly important because of the added detail and accuracy they provide.

Monitoring needn’t to be confined to electricity. Temperature, pressure, humidity, asset tracking, and access control, are some of the other parameters that are measured in data centers. As of today, it is my impression that the different systems are still fairly segregated; air temperature information feeds into the cooling controls on the facilities side, while server utilization data is the domain of IT managers. However, as pressure to cut energy cost becomes more intense, the gulf between the facilities and IT responsibilities in the data center may lessen.

However, monitoring presents its own challenges because of the volume of information that is produced. Each rack of servers can have dozens of sensors. Large data centers can have hundreds of racks. Multiply this by reading intervals every second or less and the number of data points can get very large, very quickly.

The mass of data that is produced by monitoring has given rise to a whole new industry called DCIM (which stands for “Data Center Infrastructure Management”, and is sometimes pronounced as “dee-sim”). DCIM software basically pulls together all the various nodes in the data center and presents data in a manageable format which data center operators can use to make actionable decisions. If you like virtual dashboards, readouts, meters… then perhaps a career as a data center manager is for you! Companies active in DCIM were well represented at this year’s AFCOM exposition. I met with FieldView Solutions, RF Code, and Power Assure, which are all promoting their different flavor of DCIM software to help bring order out of chaos. As an emerging market, DCIM has a healthy mix of industry heavy-weights, specialists, and startups.

The rise of sensors and analytical software in data centers reminds me of the transition to electronic fuel injection in automobiles. In order to control the operation of an engine (or data center) more precisely, a network of sensors delivers a continuous stream of information; which is then fed to controllers so the operation can constantly be tweaked and optimized. Volumes of historical information can also help to diagnose any problems when they arise (or if you forgot to screw your gas cap). In the data center, optimization can bring huge returns by saving electricity or by freeing up stranded capacity.

Government programs such as Energy Star in the US, or the European Code of Conduct are also providing a noticeable “nudge” to the data center industry. To me, the message is that data centers should move towards monitoring and reducing energy use on a voluntary basis. If not, the hand of regulation looms. This alone can be a compelling driver for monitoring products and services in the coming years. 

Share this Blog