__ __  __
   __/ // /_/ /___ _____  ____ _   __  ___      ____  __
  /_  _  __/ / __ `/ __ \/ __ `/  / / / / | /| / / / / /
 /_  _  __/ / /_/ / / / / /_/ /  / /_/ /| |/ |/ / /_/ /
  /_//_/ /_/\__,_/_/ /_/\__, /   \__,_/ |__/|__/\__,_/
                       /____/

High level approach

The design of this “BGP” router is based on an internal radix tree structure containing the routing database. Routes are indexed in the tree based on the bit pattern of their subnet, and route lookup walks the tree to find the most specific route node corresponding to the requested destination address. The router itself is based on Racket’s built-in I/O multiplexing to automatically manage I/O between multiple sockets.

Challenges

• No existing radix tree package in Racket. Due to Racket’s unfortunate lack of adoption in the networks field, nobody has written any public radix tree library for Racket that we could find, so we had to implement our own. This went through approximately 3 iterations and it’s still mildly spaghetti code.
• No stdlib SOCK_SEQPACKET support. There’s no support for SOCK_SEQPACKET in the standard UNIX domain socket module, so we pulled the source code for this module and added a mode for SOCK_SEQPACKET.
• Interesting simulator dependencies. The provided simulator has a hard dependency on Python 3.6, which neither of us actually had installed or were capable of installing, and it did not work with Python 3.8.

Testing

For the milestone submission, we included some basic internal unit tests but there isn’t a lot of code coverage on those yet. We also ran the simulator milestone tests against the current router implementation and both milestone tests are passing.