Open-source Google Cloud Run alternatives

OpenFaaS is a serverless platform that lets developers run event-driven functions and microservices on Kubernetes with little boilerplate code. Your code or a ready-made binary goes into an OCI-compatible container image, and you receive a scalable HTTP endpoint that auto-scales and exposes metrics.

Highlights include ease of use through a UI portal and a one-click install, and the ability to write services and functions in many languages. The project integrates with Prometheus for metrics and with GitOps workflows, and its topics place it in the functions-as-a-service and PaaS categories. The gateway and tooling are written in Go.

Licensing needs attention. The README states that the Community Edition is licensed for individual exploration and hobbyist use, with a 60-day limit for commercial proofs of concept, and that commercial and production use calls for the paid OpenFaaS Standard or Enterprise editions. GitHub reports the license as 'Other'. You run it on your own Kubernetes cluster.

Tau is the open-source core of Taubyte, described as a fullstack workspace for humans and machines. It is a cloud platform written in Go that spans the development environment, building applications, and deploying and running them. Topics place it in serverless computing, WebAssembly, peer-to-peer distributed systems, CDN, and cloud-native platform-as-a-service. It is released under the BSD-3-Clause license.

The quickest way to try it is to start a local cloud with the dream tool. For real use, you deploy Taubyte to your own servers or VMs, either manually or with the automated Spore Drive tool, and the documentation at tau.how covers the platform, deployment, and getting started. A video series called the Hitchhiker's Guide to Taubyte provides guided demos.

The project welcomes contributions through a standard fork, branch, test, and pull request workflow, with good first issues labeled. Tau suits developers and platform teams that want to run their own cloud-like environment, avoid dependence on a single hyperscaler, and experiment with a WebAssembly-based serverless model.