Tip 1: Ideal scenarios to deploy an N+1 UPS configuration
When it comes to the UPS, it is not enough to have redundancy built into just a single system. In India, many data centers still use legacy systems. N+1 UPS can be achieved in such systems using static transfer switches (STS). STS transfers this load to a bypass source in case of loss of power at the UPS' output. STS can be deployed in front of a single input load, such that two independent power supplies from two clusters of UPS can be used to further boost system reliability.
Tip 2: Incorporating system efficiency in N+1 UPS design
Blade servers have high power density, many a time escalating power requirements. For example, if a normal server setup requires 5 kW, blade servers may push the requirements to 20 kW. Hence you should use a higher power density UPS for blade servers.
In N+1 redundancy, the UPS is being used to only 50% of its efficiency. In normal situations, you can push it to 75%. The part load efficiency of any system is always inferior to its full load efficiency. Thus, one of the challenges
Tip 3: Things to consider during UPS module selection
Cost is always top of the mind when it comes to selection of UPS modules for an N+1 UPS setup. Apart from cost, electricity consumption is another area which directly affects UPS module selection.
In Indian conditions, if the UPS can work on a wider range of input voltages, it does not consume battery power. UPS batteries tend to wear out earlier with more usage, bringing in huge replacement costs. Thus, you should look at UPS technologies like advanced battery management, which closely monitors battery charging. If you can make the UPS work over a wider input range, then batteries can be discarded altogether.
Till date, data center managers have always played it very conservative when it comes to N+1 UPS design. They typically size the UPS according to the available connected load.
Servers are typically designed for a certain power rating. Hence IT managers should judiciously consider that adversity factor, as well as the simultaneity factor (how many loads will come at a given point of time) during N+1 UPS design. This will help them in terms of not oversizing the UPS.
Another belief is that you can increase the UPS' reliability through UPS oversizing. Such cases always involve restricting the amount of extractable UPS reliability.
It is best to go for a modular UPS system when it comes to N+1 UPS setups. In this case, another UPS module can take over if a module becomes faulty. However, in such cases you should also look at the system's hot swappability.
Tip 5: Handling UPS power consumption
The N+1 UPS' power consumption will depend on the server's kilowatt demand. Unfortunately, we continue have a mix of legacy systems and new blade servers in India.
In the future, there will be more IT equipment with power factor correction. In a conventional system, if the UPS develops a fault, it gets raw power through the bypass. Thus, continuity of power is maintained. Modular systems have an extra module, so if there is a failure, the other module still remains in online mode. Thus, there is no question of the UPS system going into the bypass.
About the author: Robin Roy is the product manager for UPS solutions at Delta Energy Systems (I) Pvt. Ltd. He is responsible for introduction of new products for different market segments, business verticals and applications, with state-of-the-art technology for power conditioning.
(As told to Jasmine Desai.)
This was first published in November 2009