DENVER — The promise of the edge is all around us, or maybe it just seems that way at the OpenStack Foundation’s (OSF) Open Infrastructure Summit. And in discussions during an update for the Foundation’s Edge Computing Working Group, panelists spoke with anticipation of deploying 10,000 edge sites and the possibilities that brings.
But, it’s not all hype and excitement as it seems there is still a lot of work to be done in operationalizing and managing these edge locations (especially if there will be 10,000 of them).
Installation is the least important, said Beth Cohen, SDN product strategist at Verizon. “Operational lifecycle is the most important — you can slam an environment on a box, but then you have to manage and operate,” she said, adding that these considerations are 99% of the edge “problem.”
Ildiko Vancsa, ecosystem technical lead at OpenStack and moderator of the panel, did note that StarlingX, which the foundation launched in October 2018, is looking into some aspects of edge deployment and lifecycle management. And she noted that the Edge Working Group has started discussions on the operational side including configuration hosts, service and management, tooling, and even integrating some components of Airship to help with this.
In fact, listed on the working group’s wiki page are a number of the challenges that the group is facing. This includes lifecycle, image, network, and storage management as well as administrative tools to provide user interfaces — each of which are followed by a note that there are projects to come.
Something that the panel debated, and reached no true conclusion on, was that this is not a particularly new problem. And, as brought up by Ian Wells, distinguished engineer at Cisco, we have already seen enterprise laptops be operationalized, which is to say that operationalizing hardware can and has been done.
Cohen followed this up by adding that there is a difference between the management and operations of the edge and laptops, routers, and cell sites and that is that most of these appliances are single purpose. But hardware at the edge will have multiple purposes and this “rips open all the management tools.”
That, and hardware (and software) disaggregation is just making these management problems harder, and finding whose problem (between software and hardware vendors and telcos) it is to fix it is becoming even more difficult, Cohen said. “The working group has struggled here,” she added.
And Chris Price, president of software technology at Ericsson, offered that virtualized control management (something that AT&T has) could be a step in the right direction as companies are able to automate and see what’s happening.
However, the panel found no conclusion to this battle — of whether operationalizing the edge was a new concept and problem to solve, or whether we can modify existing tools to make them work on edge appliances.
The Growing Edge
The OSF started the Edge Working Group two years ago to identify the problems surrounding edge computing and work toward developer solutions. The group is now working on defining a set mission and objective. Currently, in its draft form the group’s mission includes referring to the systems and applications located at the outskirts of administrative domain, as close as possible to discrete data centers and developing architectures, users, and requirements for multiple industries and global constituencies to enable development achievements around the edge use groups (and there are a lot).
Already this year, a number of new edge groups have launched. In January, the Linux Foundation started LF Edge, an edge computing initiative, that will serve as an umbrella for its five of its edge projects, two of which were started when the initiative launched.
And, in February Vapor IO launched the Kinetic Edge Alliance with a number of partners to establish a critical mass for edge computing in the U.S. The Alliance is building on Vapor Io’s Kinetic Edge architecture.
ETSI Edge Computing, also in February, created its first working group to work on the deployment and implementation of MEC-based systems using standardized APIs.
The OSF itself has a number of edge-related projects including OpenStack Cyborg, Glance, Keystone, and Ironic as well as Airship, StarlingX.