Ensuring uninterrupted power supply to the data centre is as critical as firefighting or physical safety of equipment.
In this article we want to tell our readers how the data center ensures uninterrupted operation of equipment in terms of power supply redundancy.
Data centre standardization
In the CIS and the Russian Federation, in particular, there is no single standard that would allow for an objective assessment of the data center’s ability to provide certain levels of service.
Most data centers are guided by the American standard TIA-942, which contains the bulk of recommendations on the organization of data centers. Existing TIA standards (by the way, not always suitable for Russian realities) are supplemented with more modern BICSI 002 2010.
Data centers, to which these standards are applied, have additional classification according to a set of parameters, such as their size, reliability level (Tier I-Tier IV) and purpose. In terms of standardization of data centers for energy supply in Russia, there is also no single GOST, so it is advisable to apply international experience.
As mentioned above, there are four levels of data centre reliability.
At the first level it is allowed up to 28.8 hours of total downtime per year, at the fourth, highest level, where all power supply systems are fully redundant – no more than 26 minutes.
The main requirements for energy supply are reliability, quality and continuity. In particular, the data centre must be equipped with at least two fully insulated electrical systems.
Basic provisions of TIA-942 in terms of energy supply:
- Cable entry room in the basic topology – can be one or more;
- Main Distribution Area, where the central cabling cross of the data centre, routers, LAN and storage network switches are located. Horizontal cable crossovers can also be located there. For redundancy purposes, two or more MDAs can be organized in the data centre;
- A distribution point for the Horizontal Distribution Area (Horizontal Distribution Area), a common cabling subsystem in the data centre;
- Zone Distribution Area (Zone Distribution Area), the presence of which extends the possibilities of system reconfiguration;
- Equipment Distribution Area.
Summarizing all of the above, the result is that the result to be achieved with the implementation of a guaranteed power supply system is that the responsible equipment can function in the event of a main power supply failure within the time frame sufficient to switch to standby power sources.
How the continuity of power supply to the data centre is ensured
Now that the requirements for energy supply are clear, it is worth moving on to a direct examination of systems that provide continuous power supply to the data centre.
Uninterruptible power supplies
Let’s start the conversation with the “first front” system, which is activated immediately when the main power supply is lost, namely with the UPS.
It would be wrong to talk about them as a complete replacement for the main power supply: the main task of the UPS is to “feed” the data center in an emergency for a short period of time (up to an average of 20 minutes).
Before this time expires, more “long-playing” sources should be connected or the main power supply should be restored. In addition, the UPS is also responsible for protecting the UPS from mains interference and surges and maintaining the basic power parameters within the normal range.
A detailed analysis of each variant of the AHGE organization would occupy the entire space of this article and would be interesting only to a narrow circle of readers, so we will deliberately reduce the volume of “nomenclature” information.
You have only a short list of requirements for UPS in data centers:
- Reliability with regard to system recovery time.
- High efficiency under partial load.
- Support for parallel operation with power build-up and increased redundancy.
- Possibility of scaling.
- High power factors.
- Low input current harmonic distortion coefficient.
The UPS can be divided into two types, static and dynamic.
The static UPS power module has no moving parts, except for auxiliary fans, for example. Typically, they work “in a duo” with a remote control unit that provides redundancy in the event of an accident.
Static UPSs consist of a source and a battery based on the interaction of chemical elements. It is the principle of operation that determines the limited service life of a static UPS battery.
Replacement of the battery, from which even timely maintenance work will not be saved, turns out to be an impressive investment for data centre owners – the cost of the battery is almost 50% of the cost of a static UPS.