The cost of data center outages is increasing, from an estimated $5,614 per minute in 2010 to $7,908 per minute in 2013. How can operators and users reduce the danger of outages? According to Germany’s Siemens, a large part of the answer is to track changes in the facility.

Power supply and distribution, fire detection and suppression, automation and monitoring are all key to making data centers more reliable, according to Philip Krause, general manager of Siemens’ center of competence (CoC) for data centers in India.

Siemens has energy management and building technology skills, Krause told a Beijing round table, where he was joined by a DCD intelligence analyst and journalists from the data center industry. But the main thing is to integrate different approaches, he said.

Holistic approach
“We try to adopt a holistic approach. We get involved in customers’ projects at a very early stage to ensure their data centers are designed properly and run in an efficient manner,” said Krause.

He believes there are several major reasons for data center outages. One is the improper design of data centers: “Data centers consume 20-30 times of energy of office buildings, and require professional staff to get proper designs. However, some staff from the commercial real estate background or those less experienced may be in charge of data center design, which may lead to problems. For instance, if the IT rack density grows from 1kW to 30 kW, more cables will be needed, which will in turn generate more heat; and if such heat is not dissipated properly, it may lead to fire disasters.”

He also pointed out that if fire detection and suppression systems don’t meet standards or follow guidelines, they may also result in data center outages. “Fire detection and suppression systems need to be designed and maintained in a proper way. If one system fails, another system should take over smoothly for avoiding possible outages.”
 
Krause told us that the third common reason for outages is the failure to track changes of data center assets in a timely manner, which may bring difficulties to operation and maintenance work later, thus causing outages: “This is particularly true for collocation service providers in China. Their data centers are growing very rapidly; however, their management procedures can’t catch up. Customers keep on moving out and moving in, yet such changes are not updated and recorded in time, which may influence late coming customers.”

To avoid this, he recommends automation technologies and interconnected systems to ensure efficient operation, and the idea of  life cycle management: “If you have a luxury car which runs very fast, but do not maintain it regularly such as filling gas and testing the tire pressure, you can’t expect it to run as fast as it does at the very beginning. Similarly, if you build a data center but do not maintain it properly, many performance indexes such as PUE will deteriorate.”
 
Siemens center of competence in India offers lifecycle services to the Asia Pacific region. “We have similar setup in Europe and the US, both of which are aimed to get global trends and innovations locally delivered.”

A pitch for DCIM
This is, of course, a pitch for data center infrastructure management (DCIM), currently a somewhat-maligned discipline widely seen as over-sold.

Krause emphaised the importance of power monitoring: “In case of power fluctuations or when there is a power quality problem, you need intelligent processes to switch off devices in a certain order for maintaining maximum uptime of data centers.”

He described Siemens DCIM product, Datacenter Clarity LC, as a “strong tool” for lifecycle management, which includes plenty of features that are not available in many competing products: “By utilizing CFD and virtualized 3D data center modeling technology, this software offers real-time energy management for data center rooms, including monitoring the temperature and humidity, identify hotspots and cold spots.”

Datacenter Clarity focuses on infrastructure management, and integrates with Siemens’ building management systems, said Krause: “In addition, its open architecture can integrate more than 400 open protocols, from both IT systems and facility equipment, which makes it vendor-neutral and enables customers to have a holistic view of their data centers.”

Datacenter Clarity LC is based on Siemens’ Product Lifecycle Management (PLM) software, which has been in use for 20 years. Over nine million PLM licenses have been issued, and customers include NASA who used the software to digitally design, simulate, analyze and assemble the entire Curiosity Rover currently exploring Mars.

More conventional data center projects using PLM’s data center descendant include the Swiss National Supercomputing Center, the Italian Ferrera Erbognone Tier IV data center and Siemens’ own data center in Beijing.

In China, such solutions have been adopted by telecommunication players, large internet giants including Baidu, Alibaba Tencent, and banks and other financial institutions.

This article was adapted from our Chinese site for www.datacenterdynamics.com, with editing by Peter Judge.