Name Plate Ratings Burning Holes In Datacentre Budgets

Wake up and smell the coffee. Name Plate ratings could be burning holes in data centre budgets. It’s time to move over to capacity planning based PDUs and capture surplus or wasted power.

Today, the power requirements and capacity planning of a data centre is done based on the ‘Name Plate.’ The fundamental point that assists the technology team to determine this is the name plate rating of the various devices that are to be deployed in the data centre. All equipment i.e. servers, storage devices, routers etc, have a certain name plate rating, which determines the amount of power it is expected to consume and provides the base for calculation of raw power coming into the data centre. This normally leads to excess capacity planning.

Let us see how…

Let’s assume that you want to set up a data centre with a minimum of 100 servers with each requiring 1 KVA of power. This would mean 100 KVA of power coming into the data centre. With 100 servers utilising 100 KVA of power, the next step would be to install an adequately sized UPS. In such a scenario a minimum of 100 KVA of UPS would be essential and would generally be powered up to 120/ 150 KVA spare capacity. An additional back-up UPS would also be required. In addition to this adequate air-conditioning to cool the 100 servers is needed. Now, if you calculate the total amount we are looking at xx power to run the 100 servers. While the nameplate rating of all these devices make for understanding the power load requirement in a data centre, they do not give an adequate idea of the true consumption.

More from Biztech

Future Group - Reliance Retail Deal approved by CCI RBI ban on cryptocurrencies takes effect; prohibition could force investors to tap the black market

At Raritan, we have conducted detailed studies on the power consumption in a data centre over a reasonable length of time. These studies show that average power consumption in a data centre is only 38-40% of the rated power or the nameplate power. This implies that if a data centre has deployed 100 KVA of power for servers, it is actually using only about 38-40 KVA and approximately 60 KVA is surplus and unutilized. Approximately, the same 38-40KVA would also be utilised by the UPS and air-conditioning. One may argue that this is the average power rating and that there would be peaks in server power draw. The peak average draw in server power is 48-50% across cross-functional data centres (including types and sizes). This clearly demonstrates that there is 50% excess capacity in terms of power including the already excess capacity for UPS, air conditioning and auxiliary units that have been deployed.

Therefore, it is important that we understand the true consumption pattern of power of all the devices in our data centers. The question is how?

Exit traditional IP-PDUs and enter Intelligent PDUs. These PDUs along with their management software are capable of measuring power, energy trends, peak draw etc etc at the device level, rack level, data center level etc etc. Once the above tools are deployed, data centre power consumption patters can be obtained, which would provide insights into spare capacity.

Hence when additional devices are added, one need not run around trying to organise more Power, UPS, Genset.
_
The author is Regional Manager – India and Middle East, Raritan_