This website is served up by an instance of Node.js. Node.js is a webserver with an event oriented architecture that allows it to handle larger numbers of concurrent connections than some other architectures. However, because it is single threaded, each Node process can only use up to 1 cpu core.
So, I’m thinking about trying to modify the http-server package that I forked on github to enable clustering. Clustering allows Node to start multiple processes to handle the same workload. So, instead of having 1 Node process to serve this website, there would be 1 per CPU core and each worker would be responsible for serving out a subset of the incoming requests.