Last week we saw yet another 5G carrier sign with a hyperscaler "dragon." Bell Canada announced a deal to deploy 5G multi-access edge computing on AWS Wavelength. This follows a slate of other announcements: Verizon extended its relationship with AWS in April, AT&T is working with both Microsoft and Google, Telefonica said in May it would partner with Microsoft Azure, and of course, there was Dish’s big AWS splash in April.
This seems like a strange trend given than over the last decade these hyperscalers have been considered the carriers’ stealth-competitors. The dragons provide OTT services designed to penetrate the hearts and minds of the operators’ subscribers and have trampled traditional carrier revenues in the process.
When you look at industry growth, the carriers’ fears are understandable. Microsoft, Google and AWS have more than quadrupled their combined revenues in the past 10 years. If we compare that to Verizon and AT&T, an increase of 26% seems rather paltry. No wonder traditional carriers are afraid of getting scorched.
Not to worry though, carriers have one last solid defense: physical proximity to their customers. Or they did, until these last announcements lowered the drawbridge and invited the dragon in for a snooze.
Of course, there is technological complexity to consider. Operators have tried their hand at virtualizing VNFs on third-party infrastructure. They have wrestled with some nasty details: choosing accelerators, pinning CPUs for high performance, adapting storage for applications with i/o constraints. Now, bearing down, is a whole new world of 5G containers and microservices – applications whose performance is still an unknown entity.
Carriers have had ample time to ask themselves the question: Why are we doing this again? The holy grail of virtualization lies in the cost savings of automation, but as long as end-to-end service orchestration doesn’t work between VNF vendors, the end goal remains a mirage. Combine this with the fact that the industry seems more tolerant of operational expenses (as opposed to capex) and suddenly, sending 5G functions into the cloud is looking attractive.
Now it’s time for the carriers to make important decisions about their 5G stand-alone cores. Having computing power at the edge, either on customer premises or distributed in their own networks, is a must for low latency services. A carrier may contemplate a hybrid strategy, building its own private cloud at the edge and using a hyperscaler for core nodes. This seems like a great tradeoff – on the surface.
However, the dragons are using proprietary versions of Kubernetes. With a multi-cloud strategy, 5G deployment transparency and seamless interworking is going to be a nightmare.
This dissonance in the definition of ‘hybrid’ was underlined last November in Silicon Angle's interview with Andy Jassy, CEO of AWS. Jassy explained that AWS wants to redefine the way the industry is using the word, “We think of hybrid as including the cloud along with various other edge nodes (including on-premises data centers). ‘Edge’ and ‘hybrid’ are terms that are starting to merge together.”
When the dragon wants to renovate your castle, it’s time to sit up and take note.
The carriers face other risks in letting hyperscalers entangle themselves in the network, apart from just subsidizing the loss of their customers’ mindshare. Yes, these companies can provide infrastructure more quickly and more cost effectively, but there’s no guarantee that the savings will be passed on. Edge clouds don’t come cheap, and while it may be a relief to outsource the constant security upgrades and hardware renewals, once a carrier has signed up, extricating itself will be no easy task.
In addition, network reliability remains central. Telcos have been building ‘five-nines’ infrastructure with their traditional vendors for decades. They know that in the case of a network event the vendor will drop everything and treat it as their priority. In contrast, the commercial relationship between the carriers and the dragons is fraught with complexity. Will the carriers’ customers be prioritized when the network goes down?
Meanwhile, the big operators are eyeing with envy the deep knowledge and supporting software that hyperscalers have built to serve enterprise verticals. It must be very tempting to align with a big brand that businesses have already chosen, instead of trying to compete head-to-head with a formidable dragon.
Whether it’s to simplify the technology, whether it’s the hope of winning a bigger share of enterprise revenue or whether it’s simply chasing a marketing splash, it appears that most carriers are willing to take the risk of opening the door to let the dragon cross the moat. We’ll just hope he stays tame and in the front room, shall we?