Data centres have evolved in an ad-hoc way. Computing capacity was specified and managed by IT people. The power and cooling required for delivering and safeguarding that capacity was specified by facilities or building management specialists. The servers generated a lot of heat. Traditional air conditioning was used to reduce that heat (because when servers get too hot they stop working). Usually this meant using computer room air conditioning (CRAC) units which force cooled air down into the room.
This approach does not work well with high density computing. The scalability of High Density (HD) servers is a critical aspect of their value: they can meet the escalating need of an organisation for computing capacity without demanding new physical space, but only if the underlying infrastructure is similarly scalable.
The traditional approach to cooling computer rooms is inherently inflexible, because underfloor cool air distribution is limited by the physical capacity of the ducts, and the impact of additional periphery cooling units on the airflow around a room is in reality unpredictable.
That unpredictability makes true cooling redundancy tricky and expensive with traditional CRAC units. These issues reflect the separation of IT and facilities management, but the advent of HD has made that separation untenable. The IT team will now tend to have overall responsibility for the specification of the data centre equipment.
Power Monitoring & Management
Power monitoring and management in data centres has moved from a luxury to a necessity as heat and power densities have risen rapidly. The principal of Management by Information (MBI) states that: You can only manage what you measure. The first tenet of MBI is: The accuracy of your measured data directly impacts the quality of your decisions.
While some may already have investments in first-generation power-strip monitoring units, it is important to understand what information is required to make proper power and heat-related decisions. There may always be a tendency to look for lower quality solutions; however, choosing the right monitoring solution is vital to the goal of improving and maintaining data centre energy and carbon efficiency.
According to a recent study (P G & E), just 30% of a data centre’s budget is spent on capital expenditures, while a massive 70% is consumed by operational expenses. Using standard Computer Room Air Conditioner units 54% of its overall power consumption is spent on HVAC infrastructure, with just 38% of power used to run the servers themselves. For every 100 watts needed to power a typical server, at least another 100 watts is needed to cool it.
Power outages, the high cost of generating own power, and the demand for continuous IT operations are requiring a new look at ways to deploy the IT infrastructure without compromising security, power consumption and the cooling requirements associated with conventional data centres.
The Mothapo Systems range of controllers were specifically designed and manufactured to address the unique requirements of each installation and a suite of products is available for power monitoring and the management of power at existing telecommunication racks, ICT enclosures, data centres, server rooms, plants or other facilities.
Preventing just one power related event typically cost justifies the deployment of a Mothapo Systems controller the first time a catastrophe is avoided. Think about the cost of your data centre or server room being down for an hour, let alone a day or even a week.
How much would it cost your organisation?
Can you prevent it?
Who gets the blame if you cannot?