Until recently, the prevailing belief was that large, consolidated data centres were the best option for modern organisations to house their critical applications and data. After all, IT expertise within local communities was hard to come by and local IT resources were routinely under-utilised.
With a consolidated infrastructure, physical security was deemed easier to monitor and guarantee, and the cost and complexity of running one or a handful of dedicated facilities was significantly less than running lots of smaller ones.
Though much of this remains true, a new solution is emerging that is better configured to meet the demands of today’s businesses and consumers; one that can provide the flexibility, agility, scalability and speed they have come to expect – edge computing.
Enter the edge
Rather than a replacement of the centralised data centre model, edge computing is more a complementary addition. As we move deeper into the digital age, corporate data centre/IT infrastructure footprints will begin to include a mixture of the centralised, the regional, and the local. Though the ratio of this mixture will vary from between organisations, achieving a truly optimised data centre/IT infrastructure will require a mixture, nonetheless.
Today, many organisations already use both centralised and regionalised infrastructure resources. IT is mostly housed in one or two key data centres while regional offices rely on smaller facilities, closer in proximity. Edge computing is added to this mix to address some of the pain points that surface as digital transformation journeys gather pace.
Set to be the potentially major long-term growth area for edge computing is the rise of intelligent automation. This next-generation technology generates vast quantities of data at a local level and must be processed and actioned as quickly as possible. Indeed, with the development of technologies such as GPUs, FPGAs, and quantum computing, data will be consumed faster than it can be supplied and it is edge computing that is best placed to meet this challenge.
So, exactly how will edge computing meet this challenge?
Firstly, it’s important to understand that there are going to be different types of edge computing environments. Any infrastructure that is not housed in an organisation’s main, centralised data centre footprint already forms part of an edge computing environment.
While there are various definitions of edge computing, the broadest could read:
Any data centre and/or IT resource which brings an application closer to its point of use than the original, centralised location.
As far as edge computing applications go, they generally fall into two categories:
1. Where applications/content are brought as close to the user as possible
Imagine a business headquartered in the US that’s produced a promotional video for a new product line. This video could be located in and served from one, central data centre in the US.
However, customers trying to access it from overseas might experience poor performance, depending on how many network routes exist between the US data centre and their own location. By hosting the video in strategic, global locations that are closer to as many individual customers as possible, user experience is likely to be significantly enhanced, meaning the video delivers the impact intended.
2. Where content/data is generated locally and needs to be processed and acted upon as quickly as possible
The classic example here is the autonomous (self-driving) vehicle. If this mode of transport is to become a reality, then real-time communications will be needed across the globe’s entire transport networks. Autonomous vehicles must respond to multiple variables, including the proximity of other vehicles and pedestrians, the weather, and the time of day. They will only be safe if the IT networks, both inside and outside of the vehicles operate in real-time so they can assess conditions, report back to road-side data centres, and then receive instructions on how to proceed.
While a local edge will be required to ensure autonomous vehicle safety, more regional edges would receive other strategic information, such as volume of traffic within certain areas at certain times so speed limits could be adjusted or alternative routes identified.
Ultimately, all data generated at local and regional edges will need to be sent back to a central data centre where higher-level, longer-term decisions can be made; Are new roads needed? Do more road tolls need to be introduced or the prices of existing ones adjusted? If traffic volumes are becoming unsustainable, should we incentivise employees to work from home for part of the week?
Meeting challenges such as those identified above will require data centres located within multiple conurbations. Where they are few in number and located hundreds, even thousands of miles away from end-users, latency issues will hobble progress towards a more connected world.
The barriers to an edge revolution
Technological evolution is enabling edge computing to gather momentum at an extraordinary rate. However, the speed of edge infrastructure development to meet growing demand is still a barrier. Traditional, hard-wired networks that will provide much of the edge computing infrastructure will need to be upgraded in terms of both their topologies and speeds.
Moreover, a significant proportion of edge computing applications will rely on wireless communications with 5G playing a vital role. 5G requires new infrastructures to be built and it’s not yet known how all-pervasive it will become. In the UK, there are still many regions where 4G coverage is at best patchy and where mobile devices must even rely on 3G or GPRS.
To fully exploit 5G, a new network topology is required, including new network elements with edge computing playing a key role. Whatever the solution to the 5G build-out costs and complexities, users should be aware that despite the hype surrounding 5G, it could be some time before edge computing applications based around 5G public infrastructure becomes a reality.
Edge and the data centre
As a key component of any edge computing application or solution, data centres must provide optimised environments to house a whole new breed of IT infrastructure that underpins edge computing. The decision for organisations is whether to adapt their own facilities, access the modern facilities of a colocation provider, or even to migrate wholly to the cloud. Organisations will also need to assess whether the location of their existing data centre infrastructure is suitable by defining short and long-term objectives around their growth plans and customer satisfaction targets.
Only once this process has been completed can the number and location of data centres be decided upon. It may be that only one, centralised data centre is sufficient but the likelihood is that data centres in multiple locations proves necessary.
As building, owning, and operating a host of data centres is beyond the scope of most organisations, purchasing space within colocation facilities situated close to customers or where applications are most used will be the preferred option.
Optimising customer experience is a core objective for any business. With flexible, scalable, appropriately located data centre infrastructures that incorporate colocation, edge computing is well placed to make this objective a reality.