On 20/07/19 09:57, Michael Fox - N6MEF wrote:
The IPIP mesh may be non-standard, but it is distributed, without any single point of failure. To get between two points, the two gateways have to have IP connectivity to each other. That's it. The two end-points can troubleshoot directly.
The other thing I like about the IPIP mesh is that the routing is as good as it can possibly be, because each endpoint is connected to each other. This is an issue that is critically important for us in VK. I've had to implement my own link between my IPIP and BGP connected subnets to improve both reliability and latency of interconnectivity between them. I'm not using IPIP on the BGP connected subnet, because it's one of the 44.190.8/24 subnets, and there were objects from some parts to it being in the mesh. So instead, I made my own private arrangements to bridge them.
Being forced to route through some other host (especially if it's on the other side of the world) would seriously degrade the performance of the network as seen from here, because physics (that pesky Relativity) limits RTTs to around 200 mS at best. The round trip via UCSD was 400mS at best, if it worked at all.