S 1.56 Secondary power supply

Initiation responsibility: Head of Site Technical Services, Head of IT Section; IT Security Management

Implementation responsibility: Site Technical Services

If the computer centre has high availability requirements, then the primary power supply from the power network of a power utility company must be supplemented by safeguards aimed at providing power in an emergency. When planning this, other important aspects of the infrastructure of the building, such as emergency lighting and platforms for raising firemen, must not be forgotten.

The secondary power supply of a computer centre normally consists of a central uninterruptible power supply (UPS) for the computer centre and a standby generator unit (SGU). If local conditions and the availability requirements profile of the computer centre allow it, instead of a standby generator unit this cushion function can be fulfilled by a second feed from the power network of a second power utility.

Whereas an online UPS (see S 1.28 Local uninterruptible power supply) will bridge over fluctuations and short-term interruptions of power supply, a standby generator unit will cushion longer duration power failures.

The entire IT environment should be series connected to a central online UPS. The control electronics for this UPS must ensure that coupling on start-up of the standby generator unit and following resumption of the power utility's power supply is at the correct frequency and phase.

When sizing the emergency power generator it should be noted that the rated output of the standby unit should lie above the full-load power requirements of the computer centre. This will ensure that the emergency power generator can produce the output required, for example where several power consumers start up simultaneously.

When passing the power supply from the UPS to the standby generator unit, steps must be taken to ensure that switching takes place gradually without overloading of the SGU which would result in a subsequent restart of the UPS. The individual requirements of the IT infrastructure and the other parts of the building supplied by the SGU must be considered with the aid of an agreed policy on load management.

When sizing the batteries for the central UPS, the bridging time in the event of a power outage is the critical factor. This contains the following elements:

• Delay before network power supply is restored. Only after this standby time of 1 to 5 minutes does the SGU start up.

• Switchover time until the load is taken up by the SGU. During this time the UPS supplies all users of IT equipment with power.

• Time with reduced power consumption. If there is a drop in the battery charging capacity, power consumption should be reduced. This is achieved by disconnecting non-critical consumers from the network.

• Time with power consumption only by necessary critical consumers. Should a further drop in battery capacity occur, only the most important power sinks should continue to be supplied with electricity. At this point at the latest an automatic emergency shutdown should take place with controlled, forced shutdown of IT operations, even if reversible data losses have to be accepted as a consequence.

To maintain the protective effect of the secondary power supply, regular servicing is required.

Additional controls:

• Are the UPS and SGU maintenance intervals adhered to?

• Are the UPS and SGU put through regular test runs under realistic loadings to check that they are fully operational?

• In the case of diesel-operated emergency generators, is the tank content checked regularly?