Armada is a distributed ETCD inspired key-value store. Armada is designed to operate eiter as a standalone node, standalone cluster or in Leader - Follower mode suited for distributing data in distant locations. e.g. in different cloud regions. While Armada maintains many of ETCD features there are some notable differences:
- Armada is designed to store much larger (tens of GB) datasets and also provide iterator-like API to query large datasets.
- Armada prioritize speed and performance over some more advanced ETCD features like Watch API, or Leases.
- Armada support multiple separate keyspaces called tables which operate individually.
- Even though Armada has not yet reached the 1.0 milestone it is ready for a production use.
- There might be backward incompatible changes introduced until version 1.0, those will always be flagged in the release notes.
- Builds for tagged versions are provided in form of binaries in GH release, and Docker images.
- Tagged releases are suggested for production use, mainline builds should be used only for testing purposes.
- You need to distribute data from a single cluster to multiple follower clusters in edge locations.
- You need a local, persistent, cache within a data center and reads heavily outnumber writes.
- You need a pseudo-document store.
For guidance on installation, deployment, and administration, see the documentation page.
Armada is in active development and contributors are welcome! For guidance on development, see the page Contributing. Feel free to ask questions and engage in GitHub Discussions!