The rise of the cloud has written a new future for the data centre – and many enterprises are starting to re-evaluate how and where they will access data centre services in the years to come.

By Brendan Haskins, cloud chief technology officer at Tarsus On Demand

They are not only asking whether they should consolidate data centres or move to more efficient technologies, many are wondering whether they even need to run and own infrastructure of their own at all.

Here are a few ways that the data centre can be expected to change in the years ahead.


On-prem data centres become a niche market

Ten years ago, many people would have been sceptical that South Africa’s big four banks would ever consider shifting core workloads from their own data centres to the cloud. But today, some of the biggest financial institutions in South Africa and the rest of the world run some services and applications in the Microsoft Azure or Amazon AWS cloud.

The security, compliance and latency concerns that held back adoption have disappeared faster than even some of the more optimistic industry observers expected. The landing of local data centres by the world’s major cloud providers, in particular, has dispelled most of the worries that companies had about data sovereignty in the past.

Any company today can access scalable, reliable and secure cloud services on-demand from the hyperscale cloud providers – services of a higher quality than even the largest South African companies could build themselves. Companies recognise that they don’t need to pour investment into high-performance infrastructure they rarely need, but can instead leverage this capacity on-tap during the times of the day, month or year that they require it.

On-premise data centres occupy an increasingly niche market, for applications such as remote sites, solutions that need especially low latency, and ultrahigh security environments. Unless you run a remote mine, an oil rig, or a national intelligence agency, you probably no longer need an on-premise data centre of your own.


Data centres practically run themselves

One of the factors behind the rise of the cloud is the growing sophistication of the tools and solutions used to automate basic data centre operations. The data centre automation market is forecast to reach a value of $16.1 billion by 2023, with a compound annual growth rate of about 18% from 2018 to 2023, with a flurry of innovation underway.

Today’s automation tools enable data centre operators to run massive environments in an efficient manner, increasing uptime by enabling more proactive maintenance and management as well as lowering operating costs by reducing the need for human intervention. They also allow service providers to offer enterprise customers the ability to run virtual data centres within their infrastructure.

Enterprises take comfort from the visibility they can get into the performance of the infrastructure and their ability to manage their computing capacity remotely. This applies both to those that co-locate their own equipment within the service provider’s environment and those that purchase computing, storage and other resources on an as-a-service basis.

Advanced AI and machine learning are elevating the level of automation in the data centre and will help drive new levels of efficiency, performance and resilience in the years to come. With AI and machine learning, tasks that traditionally take people hours to do can be completed in real-time.

Data centre operators are increasingly using big data and machine learning to optimise performance, identify cost-savings and improve stability. With the help of AI, they can use data to predict faults or bottlenecks rather than using it reactively to explain an issue. And end-users will benefit from AI tools that can, for example, spin capacity up-and-down as necessary without human intervention.


Green tech takes over

We’re seeing a massive shift towards greener technologies in the data centre, and not just because data centre operators want to do their part for Mother Nature. According to Global Market Insights, the market size of green data centres will top $35 billion by 2026, driven by the rise of mega data centres and the trend of collocation.

In addition to the environmental concerns, there is a massive financial incentive for data centre operators to adopt cleaner-running and more efficient equipment. Newer server and storage units run faster and yet consume up to 50% less floor space and power than older equipment.

End-user organisations also see collocating infrastructure in data centres or purchasing infrastructure-as-a-service as a way to reduce their own carbon footprint. Shared resources are greener and cheaper than building your own data centre – for example, a single air-conditioner could meet the needs of a dozen customer’s in a service provider’s data centre.


Edge computing feeds the demand for cloud computing

The proliferation of drones, robots, mobile devices, Internet of Things sensors and connected vehicles means more and more data processing is being carried out at the edge of the network. This allows data to be crunched without going back to a central server – which is useful for applications where speed and low latency are important.

Edge computing is also valuable for reducing network congestion or for applications in areas without reliable Internet connectivity. Consider the instance of tunnelling and boring machines in a mine. Edge computing can be used to process sensor data locally so that action can be taken, if for example, a machine is overheating. Another example is an autonomous car that needs to process visual data about its surrounds so that it can avoid collisions.

Edge computing will probably be used in future mostly for applications that need operational decisions to be made in real or near-real time. But much of the data collected and acted on by edge devices will also ultimately find its way to the data centre, where it can be used to drive analytics for strategic decision making. In other words, it could fuel rather than dampen, the need for cloud storage and processing power.