Three things to remember when building a data centre
The COVID pandemic has made a drastic impact on society, not only on our health but if we look at the industrial aspects, it has pushed the ‘digital’ revolution forward, from how we have business meetings to the way we learn to operate factories.
This sudden digital shift, together with the latest developments in 5G communication, is stimulating a need for more cloud-based and online solutions, which in turn is motivating huge investments in IT infrastructure such as data centres. In this article, Hiroki Nishiyama, global manager of data centre marketing at Mitsubishi Electric Corporation, explains further about the challenges facing this important rising industry.
“The recent changes triggered by the digital shift and how we communicate, are driving huge investments in data centres. In fact, the global data centre market is expected to increase by around 8% over the next ten years compared to 2020,” commented Hiroki Nishiyama.
“If we only focus on data centre infrastructure management (DCIM), then the growth rate is even higher at 17.5%.”
These figures show that data centre investments will continue to grow, which means more data centres will be built globally, with more ‘digitally’ managed operation systems being installed.
One of the questions that needs to be considered is, “What is required in a data centre?” Hardware such as IT servers are both obvious and essential but that is not enough to keep data centres running. There are three principal points that should be considered when building data centres; sustainability, efficiency, and redundancy.
Increasing energy efficiency for cooling IT servers
Energy efficiency is a key aspect of running a data centre as they naturally consume a great amount of electricity. From powering the essential IT servers 24-7, to providing backup systems in the form of uninterrupted power supplies (UPS) and last but not least, to cooling the facility. What is often overlooked by the layperson is that IT servers produce heat. In fact, it can be a surprising amount of heat, which of course is also wasted energy, but if left unchecked, will over time contribute to a gradual degradation of the electrical components, hastening the failure of the all-important IT servers.
This is why cooling is important, not only to be executed in the most energy efficient way so as not to add to the waste, but also to keep an optimum environment to maintain the performance of the servers for as long as possible.
Nishiyama added more background: “To illustrate how much heat is generated, let´s assume a server room contained 50 servers. Together they would generate around 17-18 kilowatts of heat per hour. Imagine that´s like having 15 average heaters or a small furnace running continuously. But that´s not all, there are also additional thermal effects from the UPS, routers and switches, lighting, as well as exposure from any windows which may be present. So overall a lot of heat is being generated and potentially stored.”
Nishiyama further noted: “Actually, on average electronics devices like servers are running at 30-40 degrees Celsius, and the common wisdom is to cool the data centre to around 18-25 degrees Celsius, which is a lot of continuous cooling. So, all this heat ironically needs more power consumption to generate cooling. And that means electricity bills are just getting bigger by the day, and that is why energy efficiency is one of the highest priorities.”
There are lots of imaginative ways to reduce the heat in data centres, from pioneering work to run data centres in the sea, to taking more immediate steps of turning lights off, using LEDs where necessary, reducing windows, and improving the energy efficiency of air handling and conditioning through the use of inverters on motors that drive pumps, compressors and fans.
“Let´s consider an air handling system. Power consumption can be reduced by adjusting the amount of airflow by controlling the motor with a frequency inverter,” explained Nishiyama. “In fact, you often don´t need to be an inverter specialist to set the special parameters according to the application or load as many devices can automatically streamline motor control with auto-tuning features.”
More often newer cooling technologies such as cooling towers and air handling units for example have inverters fitted as standard, but if not, they can be relatively easy to retrofit and can quickly contribute to minimising power consumption, leading to cost efficiency and sustainability.
If cooling is so important, then another concern might come to mind: What would happen if the device cooling system failed after some years? Especially in the case of data centres, equipment failure might become critical for continuous operation, especially for data centres that might be serving financial transactions.
“There are inverters that support built-in preventive maintenance features, so this might be a solution,” suggested Nishiyama. If load characteristics are monitored in real-time, degrading performance could be detected at an early stage. Such preventive maintenance features enable operators to notice abnormalities such as filter clogging or effects from worn bearings, making maintenance easier to schedule in time and help reduce downtime.
Visualising the status of the data centre to streamline operations
Considering the issue of maintenance, visualising the status of the facilities can also help keep the data centre running continuously and reduce the workload of maintenance engineers through clear and effective communication of the current status.
“Advanced visualisation can help monitor multiple subsystems, from electricity and lighting to air conditioning and even disaster prevention systems,” Nishiyama explained. “A flexible SCADA system can help support expanding operations through to the implementation of redundant monitoring systems as the need arises.”
Nishiyama added: “With a SCADA system, such as GENESIS64 for example, the real-time status of the data centre can be monitored through graphical displays of the data from various subsystems. Alarms can even be sent via email when something goes wrong. And in the case of an equipment fault, the SCADA monitoring system can be configured to list-up the potential causes in order of probability to make troubleshooting tasks easier.”
By further utilising cloud services in combination with SCADA systems, larger volumes of data can be processed, potentially achieving integrated monitoring of multiple data centres. That becomes more important as the number of data centres being built around the world, from hyper-scale to micro data centres, increases. The use of Geo-SCADA systems will enable remote monitoring from a variety of terminals such as PCs, mobile devices, and smart glasses to help facilitate the monitoring and maintenance of these locations.
Reducing risks through redundant systems
One of the other priorities for data centres is to keep the facility running, no matter what happens. Redundancy is therefore a very important focus, which not only applies to the servers themselves, but also to the various operational systems such as power supplies, cooling or security systems since they are critical factors in maintaining the data centre environment.
“The precise control and integration of HVAC systems to maintain the humidity, temperature and other air-quality factors that affect the data centre environment, is often ‘managed’ by programmable controllers (PLCs) due to their rugged industrial specifications,” Nishiyama explained.
“To reduce risks, data centres typically install redundant cooling systems where a duplicate, standby PLC may be installed in parallel to the main PLC. And in case something goes wrong with the power source or even in emergency situations like fire in the equipment room, the control will be switched instantaneously from the main PLC to the standby PLC.”
Redundant PLC systems not only make it possible to build a highly flexible air conditioning system but can also be used to ensure the reliability of other critical equipment, such as chillers, which cannot be shut down during data centre operations.
“Traditionally such redundant PLCs could be very costly since they were built of many dedicated parts which in turn added to the maintenance burden due to their specialist nature. However, more recently, high reliability PLC systems, such as Mitsubishi Electric´s MELSEC iQ-R offer a hybrid solution where a single specialist redundant switching module can be added to a standard industrial PLC. This reduces the overall system cost and any impact on maintenance, but with no compromise on redundancy due to high quality and ruggedness of today´s PLC components,” added Nishiyama.
For redundant PLC systems, it is also essential that the main and standby PLCs are separated physically, meaning a different power source and installation location to maximise the ‘protection’. In such circumstances the main and standby PLCs should be linked by an optical fibre tracking cable to avoid problems from electrical noise, but also to have a reasonable distance apart and to reduce switching time between the two PLCs in case of emergencies.
“Using redundant PLCs really is a ´Goldilocks´ issue. You should be careful when planning the location of the two PLCs that they are not too far apart, but equally they should not be too close together either and definitely not placed in the same cabinet, or then it would defeat the purpose of the redundant system,” Nishiyama emphasised. “It´s important to target a high-speed system with a switching time of around 10 milliseconds or less to enable continuous control with high reliability.”
So, how do we build a reliable data centre system?
In this era where the demand for digital services is increasing, the pressure is also rising on providers to prevent downtime of such services, even for a short time, which makes data centre availability a critical factor. In fact, today, if one of the big ‘cloud’ providers has an outage, even for just 30 minutes, it makes headline news around the world.
“Since the demand for digital services is increasing and data centres are being built to meet that demand, it is important to remember it is not just about the IT, but also how to manage your systems and make them efficient, and sustainable,“ Nishiyama explained.
Taking small, practical steps, leveraging known technologies such as using inverters for increased energy efficiency in cooling systems, visualising the overall status of the data centre to streamline operations, or above all, reducing risks through redundant systems, are all achievable for minimum cost but with maximum returns.
Nishiyama concluded: “Working with partners that have both know-how, experience and above all an eco-system of products, solutions and partners behind them is a great way to mitigate risk and manage costs in the long run.”