Hello everyone 👋

It's been a very busy week and an incredibly busy day here, so sorry this is later than usual.

I wanted to take a moment to delve a bit further in to the background on the move away from Consul.

This all about what is known as service-based networking. Traditional infrastructure uses host-based networking – individually known and controlled devices, directly addressable and persistent in their availability. Service-based networking operations without the direct control of individual hosts. This is an emerging approach to service delivery within the cloud, and is central to the development of edge networking.

The Edge Network is designed to be able to unlock the computing capacity of low-powered and embedded devices, making use of the spare capacity that exists in the devices all around us. By definition these devices are not known hosts; rather they are devices with a primary use case that has nothing to do with edge networking. In this context the ability to understand the capability of the devices, to maintain a registry of available devices and to handle the rapid on-boarding and off-boarding of services to devices is essential.

Network telemetry in the Edge Network is provided by Agent. The definition of Stargates, Gateways and Host provides an overarching architecture to enable the use of individually contributed devices. Service management for individual devices has until now been handled by Consul, which is a centralised service management platform designed for use in the cloud. It doesn't scale in an edge context, which is why we've poured so much effort in to building out a network-native solution.

As of this morning Consul is now only used for auth and agent/host version updates. This function will be switched out soon as well, at which point the Edge Network will be a fully native service-based networking platform. And the only one in the world designed for the ground up for the edge.

_In other updates_, we've been continuing with the introduction of the new health check package, which will be used with an interval based keep-alive. This replaces the existing health stream, a method that doesn't hold up well under high load. What we've learnt about gRPC streams is that a poor connection can often lead to a connection drop at anything from an hour to a couple of days, which means it's not a reliable channel for performing health checks.

Work has continued on rebuilding the Ping package to be more reliable and to fail gracefully. This is a key component of the way we find the closest device when BGP is inaccurate.

The team started working on a path based routing system to work with the tech handling the edge.network site. We've been preparing it for the updates that Ravi has been leading, which will be live soon.