What’s Wrong with AT&T’s Silicon Valley Edge Computing Test Zone?

Peter Jarich
Peter Jarich – VP, Consumer Services and Service Provider Infrastructure

Summary Bullets:

  • AT&T announced that it is building an “Edge Computing Test Zone” in Palo Alto, Calif to support developers and other AT&T partners in rolling out a diverse set of edge applications.
  • Given AT&T’s support for edge computing, the move isn’t surprising. However, it does raise questions about the set of use cases highlighted, and a specific call-out to wireless networks as well as the lack of any reference to network slicing are disappointing.

As a member of ETSI’s Multi-Access Edge Computing group (MEC) and a prime driver of the ONF’s CORD (central office re-architected as a data center) specifications, AT&T’s interest in edge computing is no secret. Combined with a penchant for announcing its networking innovations and achievements, the carrier’s announcement of an Edge Computing Test Zone should have surprised nobody.

In very real terms, then, there’s nothing wrong with AT&T’s forthcoming “Test Zone” in Palo Alto, California. It aligns with AT&T’s interests and makes sense for any carrier planning to integrate edge computing into its network architecture in the future. It’s a good idea; getting developers engaged is critical for ensuring that they will be ready to support AT&T’s network evolution plans with compelling applications. But it also falls short in a number of fundamental ways.

  • Why the Mobile Callout? AT&T notes that, at launch, the Test Zone will leverage a 4G LTE connection, upgrading to 5G when standards and equipment are ready. Ignore the fact that LTE is actually a part of what AT&T has termed its 5G Evolution. Highlighting LTE connectivity implies edge computing is a mobile technology. To be sure, it has applications in mobile networks – but it’s more than that. That’s why CORD comes in more than just a mobile flavor and why concepts such as fog computing are applicable to IoT use cases no matter the type of connectivity.
  • Why No Slicing? While vendors continue to differ on their definition of the “5G Core,” two technologies are most commonly called out as critical: Edge computing is one, network slicing is the other. There’s a connection between them. Where slicing looks to deliver virtual networks based on the requirements of a specific customer or application, edge computing resources may be something included in a slice, particularly where the application is latency-intensive. Edge computing doesn’t need slicing, but in a very real way, slicing needs edge computing. To not see any reference to network slicing on the launch announcement, then, was disappointing.
  • Why the Missing Use Cases? With a claimed focus on AR/VR, drones and autonomous driving applications, AT&T will have its hands full. That doesn’t imply, however, that these are the only – or even most important – Edge Computing applications. IoT and network analytics. Video content processing and distribution beyond AR/VR. V2X use cases beyond autonomous driving. Public safety applications where low-latency is truly mission-critical. The list could go on and on. Regardless, it’s much broader than what AT&T called out and what it should want to consider as a part of its edge computing efforts.

Again, none of this means that AT&T’s plans are flawed or are a questionable move. But few carriers have the resources, network expertise and network R&D capabilities of AT&T.  As they all look at edge computing as a part of their future (long-term and near-term) network architectures, the messages around edge computing – what it is, what it is linked to, how mature it is – need to be understood broadly.



Leave a Reply