Meta has open-sourced Rebalancer, a high-performance C++ library with a Python interface for solving complex assignment problems. It determines the optimal placement of objects into bins while satisfying constraints and objectives. According to Meta’s engineering blog post, Rebalancer has been handling resource allocation across Meta’s infrastructure for over nine years, processing approximately 40 million placement problems per day. The release is available under the Apache 2.0 license and includes comprehensive documentation, a PyPI package, and a debugging UI called Rebalancer Explorer.

Deployment and Availability

Is it production-ready? Yes. You can install Rebalancer via pip install rebalancer, which provides v1.0.4 for Python 3.12+ with prebuilt wheels for Linux x86-64 and macOS 14+ ARM64. Additionally, .deb, .rpm, and Homebrew packages are available. However, PyPI still classifies the project as Alpha, indicating that while it is stable for Meta’s internal workloads, it may evolve before a final stable release.