The following article on cooling systems intended to handle data centres’ IT equipment at the edge of the network is authored by Jonathan Duncan, technical director, Africa at Vertiv. It is part one of a two-part article.
Jonathan Duncan, technical director, Africa at Vertiv. Image Credit: Vertiv
Digital transformation means organisations are becoming increasingly reliant on information technology to run almost every aspect of their business. The creation of more data – which needs to be processed and stored – brings with it the corresponding need for more computing power and more data centres. In many cases, these are not large, purpose-built data centres, but rather close-proximity, smaller edge facilities that share space in the same building as the rest of the business. This, in turn, can create unique challenges with regards to securing the environment and cooling the IT loads.
Edge computing can be described as the concept of having compute and storage capacity physically close to where users are generating, consuming and manipulating data. We are seeing a rise in edge computing (also known as decentralised IT) driven by such factors as the ongoing rise of the Internet of Things (IoT), which generates significant amounts of data, legal requirements, the need to consolidate data, high network costs, latency issues and network security requirements.
When the latency involved in sending data to a centralised or cloud data centre is too long and becomes an inhibitor, this drives the need for localised processing capabilities, in other words edge data centres.
However, these edge data centres must often share an area that’s already serving another purpose, and which may not have cooling systems intended to handle the IT equipment at the edge of the network.
Checking the cooling requirements
IT equipment can produce large amounts of heat on a continuous basis. Organisations must therefore take steps to ensure the proper cooling of that equipment in order to protect it and ensure its availability. Placing sensitive IT equipment into spaces designed originally for some other purpose can present challenges, especially with respect to cooling.
For example, an office building is optimised to be comfortable for its employees, while a more open space, such as a factory floor or warehouse, has its own heating and cooling requirements. In either case, the existing cooling system may not meet the stringent requirements for the proper functioning of a data centre. This is especially true for high-density IT equipment, including hyperconverged infrastructure, which can generate large amounts of heat from a relatively small space.
Controlled vs uncontrolled environments: what is the difference?
Companies are now routinely installing edge data centres in two general categories of spaces, as follows:
- Controlled office environments, which are geared towards standard comfort cooling for humans; and
- Uncontrolled environments such as manufacturing spaces, which may or may not have ambient cooling and humidity control in place.
The typical office environment uses room-based cooling systems provided by building heating, ventilation and air conditioning (HVAC) systems, or de-centralised mini-split cooling systems. Cooling capacity is calculated based on the heat load the comfort cooling system needs to handle, typically measured in watts (W) or kilowatts (kW).
A typical office HVAC system could have a cooling capacity to deal with a heat load in the range of 50–100W per square metre, or perhaps 1–2kW for an entire room. But a single rack of IT equipment may produce a heat load of 3–4kW or more. As a result, a cooling system designed for 1kW of cooling could now be asked to deal with as much as four times that capacity.
This is likely to have several repercussions:
- Employees may be uncomfortable as the comfort system struggles to maintain a target temperature.
- IT equipment (such as a server) often has thermal protection systems that trigger a shutdown if the temperature rises too high.
- Overtaxing the comfort HVAC system by requiring it to continually operate above its rated cooling capacity will drive up operating costs over the long term, and probably also cause its early failure, as it is typically not engineered for continuous operation.
Continued in part two…
Source: Vertiv
