The fuse has been burning for several years, but now, in 2021, edge computing is about to explode in enterprise adoption. The cause? Three things: the virus pandemic, 5G networking, and the emergence of data center marketplaces.
What is Edge Computing?
Before going on, let’s be clear about what exactly edge computing means. The concept itself is very simple. It’s all about carrying out data collection, processing, and storage close to where that data is generated and, generally, close to where it can be usefully consumed as well.
Not a central, stand alone platform, like a single cloud or data center. Instead, edge comprises a widespread network of sensors and computing devices.
Put another way, it is “computing close to where the action is.” That could be near a factory floor where manufacturing system sensors generate data and are controlled by systems that use that data. It could be in a self-driving car, close to where all the autonomous driving sensors are located. Or it could be in a local data center or colocation facility in a town where a company’s branch office is situated.
The key raison d’etre of edge computing facilities is that by being close to where the data is generated, latency is reduced and bandwidth consumption is minimized.
In the example of an autonomous car, it would be challenging to send the vast amounts of data collected by sensors to a remote computing facility because of bandwidth constraints and costs, and the latency involved would likely mean that the car was unable to react to events quickly enough.
Shaved Cloud Growth?
The first factor driving the explosive growth of edge computing is the global coronavirus pandemic.
Many enterprise IT departments spent a good proportion of 2020 accommodating the sudden need for remote working, with a vast expansion in the use of cloud services. A by-product of this is that many planned initiatives like investments in SD-WANs, and, in particular, investment in edge computing, had to be postponed.
One year on from the start of the pandemic, IT departments are getting back in their stride, and shelved edge computing initiatives are back on the table. That’s not to say that the growth in cloud computing will not continue to be strong, but Forrester Research reckons that edge computing initiatives will shave about 5% of cloud growth in 2021 as enterprises move some computing to the edge which would otherwise have been destined for the cloud.
To get an idea of the sheer scale of demand for edge computing capabilities, Gartner observed that back in 2018 that about 10% of enterprise-generated data was being created and processed outside a traditional centralized data center or cloud. But by 2025 the research house expects that this figure will have ballooned to 75%.
PCNs Can Help
One problem that many enterprises face when generating large amounts of data from sensors in manufacturing, distribution and other Internet connected equipment is how to connect these so that data can be offloaded and processed. That’s where 5G networks come into the story.
Public 5G networks are still in their infancy and coverage is still limited to a few large urban areas, but private cellular networks (PCNs) using 5G are available to implement now. Spending on 5G networking by telcos and enterprises is forecast to accelerate this year, jumping almost 90% to $4.1 billion, according to IBM. Much of this will be to enable machine to machine communications, including linking individual devices to edge computing facilities.
There are a number of benefits of PCNs over traditional Wi-Fi networks for sensors that can’t easily be connected to a wired network. For one thing, it is easy for a network architect to manage capacity because adding indoor base stations adds capacity linearly, without any of the worries about complications due to interference or channel allocations that you get when adding Wi-Fi access points.
It’s also less likely that large numbers of access points will be needed. That’s because one indoor 5G base station with a power rating of 1W equivalent isotropic radiated power (EIPR) can provide coverage to an enterprise space of 30,000 square feet, or possibly even more. To get that kind of coverage using Wi-Fi might require between four and as many as ten Wi-Fi APs.
PCNs also offer a good level of security because data is always encrypted. The issue of passwords is avoided because authentication is done though a SIM or sum inside each client device. That means there is no need for a RADIUS or AAA server.
5G base stations can be equipped with their own servers, and thus can provide a highly convenient place to receive and process data and send it back, without having to backhaul it to a data center further away. But of course sometimes this type of setup may not be possible or desirable, in which case data will need to be sent on for processing.
Data Center Aggregation
The entire point of edge computing is that the compute power should be somewhere as close as is practical. In many cases the traditional public cloud will be too far away, as will a head office data center. But a local edge data center of colocation facility may be suitable. The question, then, is how to find a suitable facility?
The good news is that this should become increasingly easy thanks to the rise of data center aggregators such as Inflect and Edgevana. These companies work in the same way as travel aggregators such as Expedia or Kayak, enabling enterprises to search for the compute resources they need, using filters to select for location, SLAs, security certifications and so on and discover local capacity on offer.
As these types of services proliferate over the course of 2021, finding and setting up suitable local compute resources should become fairly straightforward.
The upshot of all this is that many enterprises now have the motive, the means and the opportunity to launch or expand their edge computing initiatives. The edge computing space is finally about to get interesting.