Predicting failure risks based on big data in the cloud
Besides redundancy, the reliable operation of a data centre hinges on defined maintenance intervals. On the one hand, regular maintenance work forms the prerequisite for component or system longevity, yet on the other, represents an immense cost factor. New concepts emerging on the market reveal that the IT sector is following trends in the automotive industry. Leading car manufacturers dispensed with fixed maintenance intervals long ago, preferring sensors which determine the exact condition of brake or clutch pads. The vehicle information system lets the driver know in advance when their car needs maintenance or worn parts must be replaced. In addition, independent car repair shops are now authorised to perform maintenance on vehicles. This saves customers time and money. So why not apply the same principle to data centres?
The switch power supplies of modern IT components (servers, switches, routers etc.) generate operational leakage currents, which in large data centres amount to several amperes [A] on the protective conductor (green/yellow). Leakage currents of this nature can be measured in TN-S networks at the central earthing point. In the first place, this constitutes a risk to life and limb, because the PE conductor exercises a protective function (baseline protection). The responsible electrical manager should therefore take care to monitor these leakage currents.
Power supply leakage currents are generally pulsating currents, while residual currents take the form of classic DC residual currents caused by component wear and tear. If these DC residual currents are greater than 6 mA, they are only detected by Type B RCM (RCM: Residual Current Monitoring) sensors. However, who decides what is “normal” and what is “dangerous”?
Only few specialists possess the long experience required to correctly classify residual currents. This experience relates to both basic electrotechnical applications and special metrological characteristics in systems with high leakage currents. Both can be combined in a unique solution approach: the measurement data from the RCM sensors in a data centre are transmitted to a BSI-certified cloud application where they are compared with big data from many current analyses in other data centres. In this way, the measured values of leakage/residual currents can be evaluated and correctly interpreted by experienced measurement engineers. Parallels to use-oriented maintenance in the automotive industry can be identified here.
The cloud application sends the operator or user daily or weekly reports via email drawing their attention to possible or potential dangers. However, the measured data are not only processed with BSI certification, but are also permanently stored immune to falsification. Finally, the resultant measurement logs can also be used as proof of the safe insulation level for the overall system, thus replacing insulation measurement as part of a retest in line with EN50110-1. This measurement and evaluation process is recognised by both the German Employers' Liability Insurance Association and insurances companies in Germany.
Autor: Tilo Püschel, © Bachmann GmbH 2019
