Pex's servers ingest all the songs and videos on the internet... in real-time... on bargain-bin CPUs.

In pursuit of steep cloud discounts, Pex.com takes advantage of ephemeral "leftover" compute. These constraints make Erlang's "let it crash" philosophy both effective and economical!

In this talk, we'll cover how to design fault-tolerant systems where chaos is the norm and instability is expected. Let's explore practical patterns for distributed failure/recovery in Rust, Go, and Typescript!