We have deployed a BGP route optimization system on the Togglebox network. It continuously measures the paths your traffic takes across the internet and moves traffic off any path showing packet loss or elevated latency, before most users would notice anything was wrong.
The problem with plain BGP
BGP, the protocol that steers traffic between networks, picks routes by policy and path length. It has no idea whether a route is actually performing well. A path can be technically “up” while quietly dropping two percent of packets or adding 80 milliseconds of latency, and BGP will keep sending traffic down it because, on paper, nothing is broken. Anyone who has watched a working server suddenly feel slow from one part of the country has probably met this failure mode.
What the optimizer does
The system probes routes to destinations your traffic actually talks to, comparing loss and latency across the carriers available to us. When a path degrades, it announces a better one and traffic shifts automatically. When the troubled path recovers, traffic can shift back. The whole loop runs continuously, without waiting for a human to read a graph.
What this means for you
Nothing to configure and nothing to buy. This runs at the network level and covers all traffic in and out of our datacenters. What you should see over time is fewer unexplained slow spells: less packet loss during carrier issues, steadier latency for your users, and fewer of those support tickets that end in “it seems fine now.”
