The next time the phrase ‘we operate to industry standards for best practice’ pops up in a conversation about datacentre efficiency, it might be worth digging a little deeper to see what is meant by ‘best practice.
Over recent years, our experience suggests that not all ‘best practice’ is ‘best’ by any means. So here are four of the key areas which we believe every operator should be able to address in their ‘best practice’ approach.
Managing temperature and airflow distribution means more than simply putting the AC units in the right place (perpendicular to the hot aisles since we’re on the subject). Has the designer / installer calculated the balance of IT equipment airflow vs CRAC airflow; considered the importance of uniform static air pressure (using raised floors); and calculated the impact of vent tiles in the overall design (adjustable vent tiles throughout, and removing any next to AC units).
2. Hot and Cold Aisles
The principle of eliminating the mix of hot and cold air by employing hot aisles and cold aisles should be standard by now, but it is a principle which can be, and should be, refined and built on by (for example): eliminating gaps in rows; using longer rows; and deploying cabinet blanking panels and cable entry seals throughout. God really is in the detail where airflow is concerned.
3. Equipment Rack and Containment Design
It’s amazingly easy to destroy a great principle such as hot and cold aisles with some good old-fashioned bad practice. It might seem obvious that cabinets and racks should be arranged in an alternating row pattern (with fronts facing each other) to create those hot and cold aisles, but it’s all too common to find a very different reality.
And once you’ve got that right, you’re really into the detail. Do PDU cables only run under cold aisles?
Are perforated floor tiles located only in cold aisles? Are trays for data and telecom cabling located under hot aisles? Has the design provided adequate clearance for equipment to be mounted in racks and cabinets from the front? Are cabinets aligned with one edge along the edge of the floor tile? There’s much more, but you get the picture – this is a whole-room process, with every cabinet and every rack making a real difference.
4. Monitoring and Control
If you haven’t got sufficient monitoring and control systems in your datacentre then you simply won’t be able to make informed decisions concerning its day to day operation and performance. As a bare minimum you will need monitoring tools to capture and compare performance, temperatures and power useage (make you’re your operator is taking advantage of advances in thermal imaging as well – they can deliver an immediate and powerful insight into key issues). These tools must include capabilities such as: measuring server temperatures in real time; creating historical graphs for trends; and enabling departmental billing based on energy useage rather than RU space.
And finally, once you’ve checked what a service provider knows about best practice for efficient datacentre design, make sure you check how much they know about putting it all into practice. Because ultimately, that’s where great ideas turn into efficient data centres, reduced energy bills and a shrinking carbon footprint.