As the data centre industry becomes increasingly crucial to the everyday activities of consumers and organisations alike, there are growing opportunities to use AI to manage energy consumption and distribute workloads more efficiently while enhancing cybersecurity measures.
Driving energy efficiency
Energy efficiency is one of the most important areas of focus for data centre operators. The infrastructure supporting a data centre is where AI can have the biggest environmental impact. Intelligent control systems have been in place to drive efficiency in power and cooling infrastructure for a number of decades and the ongoing development of these systems means they are continuously getting smarter.
To achieve sufficient levels of cooling for the servers, data centres are temperature-controlled environments where the temperature needs to be maintained around the clock. This requires a lot of energy and data centre operators are challenged to continually reduce energy consumption. One way around this is to build data centres in colder climates, such as countries in the Nordic region where the ambient temperature is a lot lower.
Since not all locations will benefit from a cold climate, this presents the challenge of running the most efficient solution for the local environment and this is where vast data collection, real time data analysis and AI will bring significant gains. Furthermore, as global extreme temperatures are on the increase and as weather patterns are becoming more erratic, Intelligent cooling systems are connected to networks of sensors throughout each data centre, measuring the performance of the customers’ hardware as well as the ambient conditions within and external to the building. The cooling is intelligently modulated to suit the conditions within the data centre and as these systems become increasingly intelligent, they are working harder and faster at achieving the best possible cooling efficiency.
As the AI capabilities of these cooling systems increase, data centre environments can be controlled with more granularity, giving them better visibility and control over the complex systems. Predictive modelling will also come into play, using machine learning to forecast when and where temperature fluctuations will happen and dealing with those fluctuations as efficiently as possible. As the networks of sensors grow within each data centre. This will be particularly beneficial for colocation providers who aren’t in control of customer equipment, the customer load, or the temperature fluctuations of those loads.
Improving cyber security
Ensuring robust cyber security is a crucial challenge for data centre operators. The monitoring of data centre network traffic is one way operators can ensure resilience against cyber threats. There are robust solutions that are beginning to appear on the market that will leverage AI and machine learning to monitor network traffic intelligently by learning the typical patterns of network activity and looking for anything that falls outside of that typical behaviour. These solutions look for changes in typical traffic behaviour at specific times of the day or analyse the size of data packets that are being transmitted across networks. Any anomalies that are detected will result in the automatic shutdown of ports on the networks where those anomalies are identified.
An example scenario would be an employee transmitting large amounts of data outside of office hours. The AI built into the cyber security system would detect that as an unusual pattern and flag it as an issue within a matter of seconds, potentially shutting down the ports through which the data is being transmitted. This entire process will take place at a speed that is simply impossible for a human to match. In fact, most organisations are not effectively monitoring their network activity with a view to ensure network security, rather their response is reactive to a problem that has already occurred meaning that the damage might already be done. With this functionality now appearing on the market, AI will quickly begin to add an additional layer of robustness to data centres from a cyber security perspective.
Distributing workloads more efficiently
The distribution of workloads across a data centre operator’s different locations is managed by weighing up various factors relating to business objectives. These factors can include the cost of electricity and carbon footprint of each data centre. Deciding which data centre is most suitable for each workload helps the operator to balance the whole load across its network of locations. Where the goal is to keep electricity costs as low as possible, a workload can be migrated to a specific data centre where the electricity is lowest at that exact time.
There is a significant amount of workload management that takes place among larger data centre customers to make sure that they’re using server space in the most efficient way possible. It’s likely we’ll see AI and machine learning playing a bigger role in performing the analysis, making the decisions and transferring the data instantaneously.
Different data protection laws across different regions can be an issue when transferring data from country to country. AI will likely begin to factor those regulations into its decision-making, reducing yet another pain point in the distribution of workloads.
As AI-driven solutions continue to evolve, we will certainly see more intelligent decision-making taking place in various areas of data centre management. We’re seeing AI give data centre operators the ability to constantly improve their service offering by analysing data in real time and deliver split-second decisions that can only be performed by computer.
Energy efficiency, cyber security and workload distribution are where we can expect to see this happening soonest, but in the longer term it will be fascinating to see how AI will improve the industry in different ways.