Change of culture needed to tackle cooling resilience

    EkkoAir uses IoT and 3D gaming technology
    EkkoAir uses IoT and 3D gaming technology

    Cooling has a significant impact on data centre operation and the implications of failure can be severe – thermal risk needs to be eliminated at all costs, warn EkkoSense CEO Dean Boyle and CTO Stuart Redshaw. Louise Frampton reports 

    Earlier this year, EkkoSense published startling statistics revealing that almost eight out of 10 UK data centres are currently non-compliant with ASHRAE Thermal Guidelines. Thermal issues are among the most common causes of loss of service, yet 95% of data centres do not perform controlled cooling resilience tests. The company’s Dean Boyle and Stuart Redshaw believe there is a need for a change in culture – cooling systems are not often investigated, in stark contrast to back-up power systems.

    “If you talk to data centre operators, the thought of not periodically testing their generators would seem alien. It is part of what you do and no one would question that activity. However, most data centres would not routinely simulate failure of their cooling devices because they would be too insecure of the effect of that failure on their facility. As a result, they tend to find themselves in a position where, unless machines have to be switched off for maintenance, they are not tested or cycled on and off.,” comments

    “While they advertise they are working to a resilience limit, and have more cooling machines than they need, very few of them test this. There is a big contrast in how they manage their cooling assets compared to how they manage their power assets, despite the fact that both are critical to the operation of data centres.”

    Boyle and Redshaw point out that thermal optimisation tackles three major data centre challenges – it not only mitigates risk but also delivers significant energy benefits and frees capacity. They explain that data centres can save an average of 23% on their energy costs.

    Cooling utilisation

    In addition, instead of continually adding cooling systems, data centres can keep existing cooling systems on standby and realise additional capacity. Most data centres don’t actually know their current cooling utilisation. EkkoSense’s research found that typical cooling unit utilisation rates stand – on average – at just 34%, and organisations are spending far more than they need to on expensive data centre cooling hardware.

    Stuart Redshaw

    Research also suggests that less than 5% of data centres are actively monitoring and reporting individual rack temperatures and their compliance. The result is that they have no way of knowing if they are truly compliant with ASHRAE Thermal Guidelines.

    Sensor technology

    “Often you will find just a handful of sensors being used, on pillars across the room, or perhaps one on every 10 racks or so. This isn’t of much use – a single rack could house the most important server. Every rack is affected by the cooling system,” says Redshaw.

    “There must be sufficient granularity of data and transparency and it must be simple to use and simple to implement.”

    “There is a strong argument for telemetry, to provide monitoring for all of your racks, but the safest way to know that a cooling machine will not have a negative effect on the data centre, when it is turned off, is to have sensors measuring the temperature of all of the thermally sensitive equipment in the data centre, so that you can link the two together,” adds Boyle.

    “Under very controlled conditions, you can turn off a cooling machine, while monitoring the temperature of all the IT equipment – if there is no negative effect, it passes the test. If there is a negative effect, you can turn the cooling machine back on under very controlled conditions. This is one of the key reasons for promoting sensor technology.”

    Real-time data

    The company’s approach to thermal optimisation includes a combination of immersive 3D software, Internet of Things sensors and expert thermal services, providing the real-time cooling data required to apply space, power, cooling and other airflow inputs more intelligently.

    A specialist team of thermal and cooling experts take advantage of the company’s EkkoAir technology – an Internet of Things-enabled monitoring solution – to track data centre cooling loads in real-time. EkkoAir tracks live cooling and airflow monitoring for critical CRAC cooling duty performance, enabling the 3D visualisation of actual cooling performance across the data centre room. 

    The team can track thermal instabilities right across the data centre, enabling organisations to successfully balance their thermal profiles so that only those cooling units that need to be working are active. EkkoSense’s thermal optimisation team will initially set an energy benchmark for the data centre, ensure the operator understands how power is being used in the facility and then complete a sweep of sensors across the site, so that the temperature for each piece of thermally sensitive equipment is noted.

    “This allows us to promise our customers that, during the process of optimisation, we will not allow any piece of equipment to go over the ASHRAE temperature limits,” Boyle explains.

    Sensors are then installed throughout the site, so the process of optimisation can begin. This includes modifying the position of vents in the floor, blocking air leakage, changing flow rates on fans, adjusting set points on cooling machines and changing configurations.

    Dean Boyle

    “There is a lot of detailed work undertaken to remove the data centre hotspots, increase the temperature of some of the cold spots often found in data centres, and we will improve the energy consumption for the refrigeration machines,” comments Redshaw.

    3D simulation

    Finally, the team will simulate the worst-case failure scenario of the cooling systems, using 3D modelling.

    “3D software is an important tool for thermal optimisation. You can move around the environment and it will give you a complete sense of immersion. It is a much more natural way of working; it allows you to absorb information much faster, and ultimately it ensures that you don’t miss equipment getting too hot or too cold; it is a much better way of interfacing with a large amount of information,” Boyle continues.

    “In the future. data centres will be able to walk around with handheld devices, scan them over objects and receive live temperature, thermal, cooling, capacity, power data. We expect data centres will be ahead of other industries in adopting cutting-edge technology that enables remote, immersive interaction with a facility from anywhere in the world and see enormous potential for this market in terms of thermal optimisation,” says Redshaw.

    “By applying the real-time, rack-level machine learning data gathered by our IoT sensors to the EkkoSoft Critical spatial model, we are now moving towards the kind of real-time decision-making and scenario planning capabilities that organisations will require to transition towards true AI-managed precision data centres,” he concludes.


    Please enter your comment!
    Please enter your name here