The key to making edge computing as a service practical is the capability for network operators to deploy a fast, low-latency connection – that is, they need 5G.

Mobile edge computing has been talked about for a couple of years now as the way to bring cloud computing downstream to the local processing needs of a variety of compute intensive and privacy concerned workloads. But without 5G, is it just wishful thinking? And can network operators create a market for their implementation of edge computing as a service?

Bringing compute resources to the edge means reducing the amount of data that needs to be sent up to the cloud. Indeed, for many data-intensive applications like video surveillance/security, natural speech recognition, health/wellness related monitoring, etc. the need to process data as close to the source as possible is mission critical.

Why some data needs to stay at the edge

The reasons are multifaceted but boil down to three primary issues: keeping the networks clear of massive amounts of data needing to be sent up into the cloud, keeping latency low to allow rapid response as the data is processed locally, and privacy concerns that are alleviated (mostly) by keeping data processing as local to the data source as possible thus eliminating massive amounts of potentially hackable personal data from being stored in the cloud.

There are many ways to segment cloud computing so that a localized processing capability can be deployed. Indeed, products like Microsoft’s Azure Stack, Google Cloud IoT Edge and AWS’s IoT Greengrass are designed specifically to run in an edge environment while maintaining all of the features and functions familiar to cloud users (e.g., cloud management functions, analytics, etc.).

By running at the edge, these services can exhibit near-real-time performance with minimal latency, operate with intermittent or no wide area network connectivity, enable secured data transmissions and interactions, and run on edge computers scaled for the particular needs of the application, from small to large server implementations. This flexibility is critical to many implementations of IoT solutions like smart cities, security, mission critical healthcare, autonomous vehicles, etc.

What’s become a debate lately is just where these edge servers should be placed. Network operators (e.g., AT&T, Verizon, Vodafone), looking to take maximum advantage of their move to network function virtualization (NFV) in their core networks, which is implemented with general purpose computers as opposed to past generation of specific purpose-built devices, want to convince customers to use their systems for their edge computing needs. This is essentially offering edge computing as a service, and network operators are gearing up for this (e.g., AT&T and Microsoft announced a major partnership, as did Verizon and AWS).

In some cases, this could be an attractive use of edge computing, particularly if the scale is large and the available connectivity to the edge is fast with low latency. Such systems could prove attractive for large-scale deployments of smart cities, autonomous vehicles, medical services, transportation systems, etc. Of course the key to making this all practical is the need for network operators to deploy a fast, low latency connection – hence the critical need of 5G to implement such solutions.

But not all edge computing systems need fast speed/low latency. If the processing is not being done in real-time, or if the scope of the deployment is limited to a few servers and/or small scale device connectivity with low data requirements, the need is less apparent. So how should you decide if your company needs edge computing and if the network operators’ 5G implementations are right for them?

Workloads and processor requirements

Different workloads obviously require different amounts of processing capability. The trouble is, many companies don’t always grasp the needs of their workloads. Indeed, the attractiveness of the cloud is that resources can be scaled up or down as needed. Using the network operator’s infrastructure can bring such advantages, while implementing your own edge computing means much less flexibility. If you are firm in your belief that you fully understand the needs of your computing resources currently and into the future, then setting up you own edge computing systems may make sense. If you are not confident, then the flexibility of using the network operator’s edge system could be very attractive.

It’s also about how close to the network your workload needs to be.

Some workloads require very limited connectivity with very small data sets that only get updated on an infrequent basis (e.g., machine maintenance analytics, some environmental monitors, etc.). In this case, the need to be tied directly to a high speed network is not a requirement.

Almost any low-cost (but reliable) network will do and there is no real need to require the speed and latency of the network operator’s 5G network. However, one of the major features of 5G is that of network slicing — allowing operators to provide a reduced portion of their network for just such needs at a significantly reduced cost. As such it may still be attractive to work with the network operator’s edge systems.

It’s all about cost/performance

Can the operators provide a cost-effective option to placing your own servers at the edge? Management and maintenance of the edge servers falls to the network operator if you choose to implement your workloads on their edge devices. This can be an attractive option for those who want to keep capex spend to a minimum, or who don’t wish to invest in their own IT resources for the maintenance and operation of their own devices. It’s also attractive if your plans could change quickly and therefore don’t want to be burdened with fixed equipment that could become obsolete.

Bottom Line: While 5G holds the promise of significantly enabling more mobile edge computing systems than currently available, it is not necessarily the answer for all organizations. However, the availability of network carrier’s edge services provides a needed option that significantly enhances the capability for companies to deploy edge computing. But organizations must assess workload needs, amount of data generated, necessary speed/latency of processing, and privacy needs in order to best analyze whether a 5G-enabled implementation of a network operator’s edge computing as a service solution is optimum.