The use of digital twins is a design method already implemented by the likes of Google and Tesla when it comes to car design, but how can this technology be translated to the data centre? Jonathan Leppard, director at Future Facilities, takes a closer look.
Fully autonomous vehicles are on their way. Already in parts of sunny California and Silicon Valley, cars from companies such as Tesla, Google and BMW are driving around on public roads with no human input. Following the recent news that the Department of Transport is updating its rules and guidance on autonomous cars, we will soon see them on UK roads too.
However, while there has been a lot of talk about LIDAR (Light, Detection and Ranging) and5G technologies, one key area has been overlooked, that of the data centre. This is a major oversight as data centres will need to be rethought if they are to enable autonomous cars.
Edge data centres in particular, bolstered by 5G’s incredibly low latency, are set for huge expansions as we look to support these emerging technologies. This rise in usage and increased computing in edge data centres will need careful management of the huge levels of heat that will be generated if additional downtimes are to be avoided.
Interestingly, to manage these thermal issues correctly, data centre designs are increasingly turning towards a design method already used by the likes of Google and Tesla for their car design: the digital twin.
Digital twins and connected cars
A digital twin, in the simplest sense, is a virtual replica of a real-life installation or scenario that uses data to suggest any changes that are needed. The advantage of a digital twin is that ideas, concepts and solutions can be safely tested, changed and retested very quickly and easily by designers before an update is brought into the physical world.
In the connected car industry, companies such as Nvidiaare already using digital twin simulations to train self-driving cars and help them learn difficult road layouts in a virtual world. By repeating training scenarios multiple times in these virtual environments, cars can quickly learn how to deal with difficult situations before deploying a real vehicle on a real street.
Using a similar methodology, data centre designers can use Computational Fluid Dynamics (CFD) to power digital twin simulations of a data centre. Through testing in a virtual environment, designers can quickly test, hone designs and gather data that can inform decisions on creating the most optimum and efficient layout. With the rise of edge data centre computing and the need to very carefully manage heat generation, CFD-powered digital twins are playing a crucial role in ensuring that connected cars have access to the computing power that they need.
Creating a digital twin of a data centre means that highly accurate predictions can be made regarding any changes and updates in the design. By using CFD to model the flow of the air, or water, that is cooling the data centre systems, designers and engineers can develop layouts that maximise the space available; all whilst reducing the risk of any downtime or an outage due to thermal complications.
The impact of 5G on airflow and simulation
Looking to the near future, the need for CFD-powered digital twins will become even more important as 5G usage increases. The growth in data and processing taking place outside of centralised data centres is huge. Gartnercurrently estimates that around 10% of all enterprise data is generated outside of a centralised data centre; however, this number is set to grow to 75% in the next six years. This rapid change will no doubt send power densities soaring as we cram more compute into smaller edge sites, which in turn will put a huge strain on the cooling systems in place.
Failure to provide effective cooling will jeopardise the success of 5G technologies from their very inception. We will certainly need to do better than the average utilisation of current facilities which is around seventy percent.
For instance, while connected cars are one very obvious example of how edge data centres will be used, there are a number of other applications too. In fact, the recent announcement of Google’s Stadia suggests that gaming could soon benefit from these capabilities.
With game consoles being run virtually in an edge data centre, issues such as latency, speed and power can be negated. As even more edge computing use cases emerge, demands on the computing power of data centres increase, leading to a huge spike in the level of heat generated. A digital twin then can be used by data centre designers to predict how cooling can be changed to limit or remove any increases in heat in the most efficient way.
By testing out different layouts in a digital twin, designers can safely arrive at the best layout before implementing it in the real world. In addition to this, there’s also the benefit of improved communications between the designers and future operations teams. In fact, for the first time, design teams will be able to hand over valuable data which will enable operations teams to better achieve the performance level it was originally designed for.
Designing the future
The use of digital twins is a huge benefit to designers. Already, those developing autonomous cars are using them to train vehicles to understand the most complex of road layouts. However, the use of a digital twin will not stop there. For 5G and IoT technologies to bring us into the future, then data centre designers will also need to leverage them too. So, while 5G and IoT technologies will continue to draw attention and headlines, it is vital that we do not overlook the changes to infrastructure that will be needed to power our autonomous, connected future.
