Trends in data centres in 2018: What matters to IT managers

By Rittal product manager IT infrastructure Clive Partridge

Digital transformation is in full swing, to the extent that at least half global value creation could be digitised by 2021, according to a forecast from IDC market researchers.

Faced with high electricity costs, it is becoming increasingly important for companies to modernise the IT landscape and make their data center operations more efficient.

The following review of managed cloud services, edge computing and direct current in the data center, analyses which are the most suitable technologies to make ongoing operations both cost-effective and future-proof.

Hybrid multi-cloud environments will dominate future IT agenda. According to IDC, more than 90 percent of companies could be using multi-cloud platforms by 2021.

There are many reasons why. For one thing, there is no one-stop cloud provider that can meet all the requirements; complete cloud stacks always come from multiple providers. Moreover, performance, latencies, compliance and risk management often have to be implemented individually, sometimes with different cloud providers.

Typical cloud services include infrastructure services (IaaS), applications (SaaS), and development platforms (PaaS).  Those who believe the sector is becoming too complex can rely on external providers to deliver managed cloud services.

Cloud systems in data centers can be operated in a completely fail-safe way and maintained by an IT service provider, while users can easily access its resources via their web browser or a desktop application. To support this shift in service provision, Rittal and its partners will increasingly be offering turnkey data centers including cloud platforms and managed services for fail-safe infrastructures.

In future, as well as expanding central data centers, many companies will be focusing more intensively on establishing decentralised IT capacities. The driving forces are, in part, modern Industry 4.0 (IoT) applications. Automated production facilities mean a large amount of sensor data has to be processed on site, in real time; data transfer to a central data center would delay real-time processing and overload networks and legacy systems.

And that's not all. Many other Internet of Things (IoT) scenarios also need extra 'edge' data centers. These include networked households and smart homes, wearable fitness trackers and smart watches, as well as networked cars and IT infrastructures in smart cities. By 2019, 40 percent of IoT data could be processed and analysed by edge IT systems, the IDC analysts say.

The new 5G mobile standard will also drastically increase the volume of data needing processing. At data transfer rates as fast as 10 GBit per second, for example, a movie can be transmitted in HD resolution in just a few seconds.

Those wanting to run IoT infrastructures as part of fast 5G networks should make sure that the required server performance is provided at an early stage so that applications can use the full network capacity. Edge data centers can be used for this purpose. They enable the rapid and decentralised establishment of IT infrastructures to supply remote production sites or smart cities quickly with more computing power on a selective basis.

But what makes an edge data center stand out? They are turnkey IT solutions, which are modular and scalable, either as racks, or complete, within containers and they are suitable for companies of all sizes. The components for cooling, power supply, monitoring and security are pre-installed and coordinated with each other; so an edge data center can be created very quickly.

That being said, central and homogenous, hyper-scale data centers are still going to be needed.

Hype-scale infrastructure is laid out for horizontal scalability to provide the highest levels of performance throughput, as well as the redundancy necessary for fault tolerance and high levels of availability.

Operators, of course, have to optimise the running costs of their centres. DC racks are one solution, improving energy efficiency. But two new IT rack standards have become established on the market in the shape of the OCP (Open Compute Project) and the Open19.

Inside the IT rack only one central power pack supplies the active IT components with DC power. This reduces the energy costs of each rack by about five percent. Certainly worthy of consideration.

Alternative energy and cooling concepts will further reduce operating costs.

Electricity from renewable energy sources, with air or seawater cooling, is one example, as used by the Lefdal Mine Data Center in Norway. This is a data center built in a former mine. It is cooled with seawater and utilises electricity generated by renewable sources. Companies can procure cloud services directly or operate their own private cloud systems.

Energy recovery is another IT cooling concept for higher efficiency. This uses the waste heat generated in the data center for building climate control purposes, in order to heat hot water. The technology itself is not new, but the aim is to develop a long-term strategy that exceeds the usual ROI calculation of three to five years.