Ericsson’s Edge Gravity quietly disappeared from view in May. The gravitational strength of the edge offerings from public cloud providers, and the increasing alignment of CSPs with them, seems to have put pay to Ericsson’s ambition for edge, beyond supporting 5G. Enter edge gravity in a search engine, and you now get a general Ericsson web page on edge computing strategy.
To win at the edge requires infrastructure investment
As we reported in Preparing for Edge Cloud, the winners at the edge will be those that have the critical mass and focus to proactively invest in edge infrastructure, and to optimize the operational cost of this infrastructure. An effective edge will require global scale and ubiquitous presence to be a compelling platform for developers and innovators. From the evidence this battle seems to be being led and won by the web scalers.
Ground hog day
In Winning at the Edge we highlighted the failure of CSPs and NEPs in creating and monetizing data centre infrastructure, losing ultimately to the web scalers. The demise of Edge Gravity, would seem to fit into this narrative, but now at the network edge. However, unlike data centres this has potentially game changing impact on the Network Equipment Providers and CSPs.
For CSPs it means that increasingly value added services (those dreaded OTT services) will be being run on web scale infrastructure at the edge. CSPs could become marginalised as providers of transport network, access network and edge locations.This is not necessarily the “dumb pipe” death sentence that many fear. Appledore in fact believe that many CSPs will increasingly flourish as utility telcos. Shorn of their need to be digital service providers they can focus at operational cost and ease of use by all. We will be producing research on the utility telco later in the year.
For network equipment vendors it could mean increasing disaggregation and commoditisation. Network functions at the end of the day are simply software functions. If you can deploy a function to a web scale edge cloud, you can deploy a network function to that web scale edge cloud. Increasingly the only barriers to moving network functions to the public cloud are historic CSP practice.
We are already seeing the movement of BSS and OSS to the public cloud, something that only a few years ago would have been seen as impossible. With public cloud at the edge, how long until, network functions make the same journey. Similarly, the distinct network centric virtualisation architectures, like MANO, could be under threat. In a future Appledore white paper, IBM noted that they are now working with a CSP, where the NEP role has been pared back to simply being a provider of VNFs. At this operator there is no MANO stack, simply software functions orchestrated onto a virtual and container based infrastructure.