By Millad Farahani, Research Director, Layer123
"Edge does not need 5G but 5G needs edge" - Sue Rudd, Director Networks and Service Platforms, Strategy Analytics Inc
We are seeing a growing focus on all things edge. Modern edge computing has come on leaps and bounds since the advent of virtualisation technology, which has catalysed deployment and initiation of a broad spectrum of applications on edge servers.
According to Gartner, by 2022, over half of enterprise-generated data will be created and computed at the network edge. The winners in this market will be those that proactively and carefully invest in creating this market. And the losers in this market will be those that wait for compelling use cases or standardisation.
With this in mind, we sought out analyst and operator opinions on what might change in 2022 regarding edge computing and how operators can maximise its potential.
What the analysts think
Francis Haysom, Principal Analyst at Appledore, believes that existing cloud economics is fundamentally based on maximising multi-tenancy of compute, whilst minimising the cost of that compute through operational scale. Network Edge Cloud, by definition, cannot have this level of operational scale with many smaller compute capabilities. Additionally, Network Edge Cloud also has less demand driving multi-tenancy as it becomes more localised. The added costs of network edge cloud need to command a premium for latency and/or communication cost reduction that is not provided by existing public cloud.
With 2021 consolidation of the edge computing ecosystem being an unlikely scenario, Tier-1 telecom operators will look to adopt a multi-prong approach to their edge strategies. Roy Chua, Founder and Principal at AvidThink, believes that due to the strong relationships that the hyperscale clouds have forged with enterprises and application developers, telcos will be pushed to foster relationships with multiple hyperscale cloud providers. Nevertheless, many Tier-1 telcos will enact a multi-cloud strategy for their mobile and fixed network edges as well as on-premises edge solutions for their enterprise customers, so they can maintain some leverage against the hyperscalers.
At the same time, says Chua, 2021 edge deployments will mostly be trials and pilots for early use cases. Ubiquitous 5G coverage and mobile edge availability will take time to roll out, so both hyperscalers and telcos alike are understandably cautious about building out too early ahead of market adoption. As use cases become validated and applications are rearchitected to take advantage of edge locations, an increase in build-out momentum is anticipated for 2022.
Johanne Mayer, Founder at NaaS Compass, goes further by suggesting that service providers need to become part of the enablement by exposing their network capabilities – including edge and 5G – to developers via APIs if they want to compete with the hyperscalers. Service providers should look at Network as a Service (NaaS) transformation to expose their network capabilities to internal IT teams and external developers. This will help them benefit from automation and the API economy.
ACG Research echoes the outlook at AvidThink, with Chief Analyst Paul Parker-Johnson stating that delivering applications at the edge provides significant new opportunities for CSPs and their customers. They are not alone in perceiving the upsides available in those opportunities and need to cultivate their offerings well to gain maximum advantages from them. In 2021 , there is likely to be a significant amount of experimentation and offer testing, with a keen eye on determining where the most attractive early adopter engagements will be found. ACG Research's expectation is that there will be 'landings' in both the enterprise and consumer market segments that have both staying power and long-term opportunity. The mix will vary greatly by provider and geography, but they will be there.
Speaking more frankly, Sue Rudd, Director Network and Service Platforms at Strategy Analytics, states unequivocally that "Edge does not need 5G but 5G needs Edge". She goes on to explain that edge services do not need to wait for 5G Service Based Architecture (SBA) to bring value closer to end users. However, cloud native 5G SBA - which will be deployed from Q3 2021 - benefits massively from pre-deployment of distributed edge compute and storage resources. The availability of low-cost resources close to end users will accelerate new 5G low latency mobile apps, secure ‘slices’ for private VPNs, video delivery buffering, and most importantly seamless fixed (WiFi, wire, cable) and mobile wireless access to shared 5G core services.
Early CPS edge services are already being hosted on AWS, Google, IBM and Microsoft Azure, says Rudd. But CSPs have a good business case to also deploy edge services for their own internal use, as a ‘critical mass’ of processor and storage platforms set the edge on P/GW, edge routers, base station controllers, CUs (Central Units) and even Distributed Units (DUs) in the next two years.
There is a possibility, Rudd believes, to deploy internal edge service platforms as a virtualised distributed system or ‘network fabric’ for video buffering, local breakout, private and real time network data backup, secure access processing, backhaul minimisation, and so on. If this happens, operators can optimise resource use and reduce CSP operations costs. Most deployments should pay for themselves in around three years and be ready to expand as 5G services escalate in 2024. Additionally, early edge resource deployment in 2021 and 2022 could allow CSPs to both partner with and ‘outflank’ the cloud hyperscalers at their enormous, climate controlled, remote metro area data centres.
What the operators think
But what is the direct view of the operators? Sweden-headquartered telecommunications operator Telia Company sees edge computing at the forefront of its agenda in 2021. Although Mattias Fridström, the company’s Chief Evangelist, does explain that as a carrier Telia will most likely not build all the way to the very edge, it will still focus on picking up traffic to bring it the long distance it needs to go. He goes on to emphasise that Telia will also move the edge more away from the core networks, meaning the edge will change slightly but not massively.
The question that is key to Telia Company is who will actually build the real edge? Will it be the current domestic service providers (mobile or fixed), or will it be more local state/regional service providers. Will it be the hyperscalers themselves? Or could it even be the equipment manufacturers with an expanded offering? Fridström puts forward the possibility that it may be completely new companies who will have a smart offering for edge to pick up the traffic and hand it over at some point to larger providers.
According to Wenyu Shen, Technology Development Manager at NTT, the biggest challenge is to find an optimal balance between latency and the cost. Although it is clear that the lower the latency, the more benefits edge power can bring, it is hard to make a decision on infrastructure investment at this early stage. In Shen’s view, one potential approach is to establish a mechanism including APIs, billing, etc. to share edge resources among service providers and even with hyperscalers. Only through this joint effort can the industry accelerate the pace of edge deployment.
Ahmed El Sawaf, Network Transformation and Cloud Infrastructure Technical Director at Saudi Telecom, builds upon Shen’s points by stating that edge is a monetisation engine of the cloud – or ‘cloud with some constraints’. He believes most operators are exploring the edge, with some spending time on defining edge classification and location. The most important aspect is a well-defined use case, as without this no budget will be allocated.
El Sawaf recommends that operators define a complete value chain of the edge – embracing a complete ecosystem that can help accelerate edge realisation. He also stresses the importance of exploring the dilemma between public and private edge, which should be ingrained within the operator's strategy.
To conclude, 2021 will see edge computing take centre stage. Operators and enterprise companies alike need to accept the participation of hyperscalers in regard to edge offerings – maybe treating it as an opportunity rather than a hindrance.