Keysource Blog

The Next Generation of Data Centre Topology

Nov 29, 2018 1:51:58 PM

This was first published in the November issue of Digitalisation World. You can read the full magazine here


Data is already in demand and the rollout of the new 5G network will precipitate a massive change in IT infrastructure requirements, according to Ted Pulfer, enterprise consultant at critical environment specialist Keysource.

The hype when 4G first launched was driven by a huge boom in consumer demand for faster mobile networks. At the time, social media had reached the masses and using your phone as a tool for entertainment was becoming increasingly popular. The sheer volume of devices streaming data created a need for faster mobile networks and I think it’s fair to say most people could see the benefit of having faster connection speeds at the time.

The next generation, 5G, is a little different. 4G is well established and speeds of 50 M/bit/s work for most people. No doubt there will be people who ask whether a new cellular roll-out is really that important.

However, as with many leaps in technology, people don’t necessarily realise whether or not they need something until they experience it. A superfast mobile network could be a catalyst for significant change in how people access data and information – both at work and at home – through mainstream adoption of IoT, AI and VR. The revision of the topology of data centres needs to happen at the same time and we need to get serious about edge-computing before 5G outpaces the industry. To do that we need to resolve some underlying challenges the data centre sector has faced for years.


Next generation

In October Vodafone became the first British network to offer UK-wide 5G functionality after it won the largest chunk of the 5G spectrum at this year’s auction. The trial is seeing the telecoms giant stream mobile data traffic to and from the internet exclusively over 5G from a site in Salford, Greater Manchester, which is connected to Vodafone’s nationwide converged fibre network. Other providers are also in testing phases in smaller locations and, with Vodafone’s push, they won’t be far behind in their own roll-outs. By 2019, 5G-enabled handsets are expected to be on the market after which point 4G will feel sluggish by comparison.

Vodafone has already conducted the UK’s first holographic call using 5G spectrum from its Manchester office with Steph Houghton, England and Manchester City women’s football captain. When 5G is all around us, consumers will be able to play 4K resolution games on the go with little to no lag and augmented reality will become even more seamless on mobile.

Now, of course, the things that 5G promises can’t be simply streamed from a single cloud source when you factor in how many devices will be accessing data at the same time. And I don’t just mean smartphones. Potentially dangerous latency issues for driverless cars, for example, can only be resolved through a change to the data centre landscape. In fact, it’s only through an overhaul of the cloud model that the industry will be able to carve out a route for businesses and consumers to enjoy 5G.


Redistributing data

Edge-of-network computing, where data centres themselves are multiplied and distributed to act as a localised portal for connectivity, will be the next trend in cloud. Data storage is already shifting away from individual devices almost entirely and cloud infrastructures as we know them are becoming vessels for data alone – given the increasing amount of data processed daily.

Edge computing is the necessary element in the middle. In simple terms it’s a series of data centres distributed across a broad geography that can bring applications physically closer to users to reduce lag. Packets of data are processed locally and it’s only anomalies that are sent to the cloud. The arrival of 5G will make this whole process quicker, especially in remote locations.

Many people accept this coming boom in edge estates. Yet, questions over the delivery have so far hindered edge deployments.


The competitive edge

The difficulty in deploying an edge-of-network infrastructure is that it’s an entirely different beast to traditional cloud structures.

One of the major factors to be considered is security. Cyber-attacks are increasing in both scale and frequency, while problems originating from physical infrastructure have also been found to be to blame for significant outages. According to the European Commission, cyber-crime is costing €265 billion a year and some experts have predicted that edge computing potentially represents a weak point for cyber security. These concerns will rightly mean that clients will expect data centre operators to be investing heavily in security and disaster recovery processes as well as the physical security and maintenance of these localised data centres.

The availability of utility services will also be a pressure point. As we’ve seen data centre requirements grow around the world, it’s a very real concern that energy demands will outgrow available supply soon. With distributed computing we’re talking about thousands of servers spread over a huge geographic footprint. Having to deploy multiple smaller data centres in this way will undoubtedly cause further strains on current power networks.

As a result, edge deployment will need to be designed with power in mind and include waste heat or localised renewable power generation.

The major question that clients will want answers to is how these data centres will be serviced with the above in mind. It’s a good question, and one that many data centre consultants and FM providers are eager to evade. The reality is that this is the major change that the entire sector will need to get to grips with. The industry has a skills-shortage that’s already causing challenges across the sector and the emerging data centre model will need greater on-the-ground maintenance over a larger geographic area. We’re talking about work volumes that aren’t currently achievable.

A long-term solution is to encourage new blood to enter the industry. Keysource, like others, is drawing in new talent through our graduate and apprenticeship initiatives. But this will take time and so smarter solutions are required. Unmanned, or “dark” facilities are one potential avenue. This is where AI and machine learning support in the management and maintenance of data centre estates, reducing the skill level for maintenance locally and reducing costs for clients too – helping them stay competitive.

5G is rapidly approaching and distributed computing landscapes have been widely accepted as the best data centre format to support it. We need to get beyond just accepting this and start acting now for how it is going to be delivered. It’s an exciting development with huge opportunities but we’ll risk delivering edge too quickly and too expensively without preparing our approach to power, security and service now. As a result, edge delivery is something Keysource is actively pursuing – modelling various delivery options to help clients embark on the process sooner rather than later.

Topics: Insight

Ted Pulfer

Written by Ted Pulfer