Author: Mark Seymour, Founder and Chief Technology Officer, Future Facilities
Choosing and operating data centre cooling equipment is a daunting task for the non-expert.
Most public conversations talk endlessly about energy efficiency and this is reflected by Power Usage Effectiveness (PUE - a measure of how much energy is required to run the data centre compared with the energy required to run the IT equipment) being the only metric in common use by operational data centres.
You’d be forgiven then for looking for the most efficient components.
This may be a good thing to do, but in many cases you’ll probably be better off if you start from a different place as a good design, operated well, will almost certainly deliver an efficient solution. On the other hand, selecting efficient components is no guarantee of an effective system.
Rather than starting with equipment technical performance, you’re better off asking some fundamental business questions around what is needed.
Questions like “How resilient does the data centre need to be?” What is the appropriate business balance view between energy efficiency and cooling effectiveness?
Once you have these concepts thought out, you are in a position to consider cooling options.
For example, in the UK, if you don’t need the very highest levels of resilience whatever the cost and your power densities don’t demand liquid cooling, then direct or indirect fresh air cooling are options worth considering.
Whatever you choose, while a single cooling unit should perform as the manufacturer specifies, once you place multiple systems adjacent to each other, on or around the building, their performance can degrade.
As a result, the proposed configuration performance needs to be tested to avoid disappointment at some later date when you find the promised cooling is not available.
The concept of configuration is not just a design issue.
How you configure the data centre is critical. For example, in an upgrade don’t assume a replacement cooling unit will give better performance because its capacity is higher.
It may have a different airflow/kW that isn’t appropriate for how you operate your IT – this can result in an airflow shortage even though kWs cooling are available.
Similarly, internal cooling performance can be affected by configuration changes.
Examples include IT deployments causing bypass or recirculation and failure or maintenance operation resulting in inappropriate cooling delivery.
Overall though, there is one overriding factor to be aware of.
To get the capacity and performance you invested in when building your data centre, you need to manage the performance with an ongoing capacity planning strategy that is applied throughout the life of that data centre.
Only then can you be sure you have got your data centre cooling sorted.
Using tools such as a digital twin can go a long way to making this process easier, quicker and clearer to manage with the ability of bringing together all the key stakeholders.