Vertx

This is a cross-post. I wrote this up for the dev.to audience. For the full architecture, benchmarks, and SDK compatibility numbers, read the in-depth write-up here on the site: Introducing Floci. Floci’s ~24 ms startup is not one clever trick. It is four battle-tested open-source projects stacked on top of each other, each doing the job it does best: Netty for high-performance async I/O. Vert.x for the reactive, event-driven toolkit. Quarkus for the application framework and build-time optimization. GraalVM Mandrel to compile the whole thing to a native binary. Together they take startup from the 2 to 6 seconds a typical JVM app needs down to ~24 ms, and idle memory from a few hundred MiB down to ~13 MiB. The honest takeaway is that this speed is borrowed: years of production hardening in those libraries are what let Floci spend its own effort on the part that matters, AWS protocol fidelity across 45 services and 1,925 SDK compatibility tests (Java, Python, Node.js, Go, and Rust). ...

Talk given at ITLA Santiago, September 2018 What Are Reactive Systems? Modern applications demand a new architecture. Reactive systems are designed to be: More flexible and adaptable to change Loosely coupled: components interact without tight dependencies Scalable: from a single node to thousands Fault tolerant: they handle failures gracefully instead of crashing Highly responsive: they respond in a timely manner, always The Reactive Manifesto The Reactive Manifesto defines four core traits that every reactive system must have: ...