MLambda.Data
A distributed database in which every component is an actor — and the consistency model is a checked TLA+ specification.
Status: in development
What it is
An actor-native distributed database on the MLambda.Actors runtime. Storage, indexes, filesystem, query planning and authorisation are all actors; there is no shared mutable state and there are no mutexes in the data path. Correctness comes from message ordering, not from locks.
What is built
A log-structured storage engine — write-ahead log with CRC32 integrity, sorted in-memory memtable, immutable SSTables, throttled background compaction. Indexes for sorted range queries, constant-time point lookups, probabilistic existence checks and full-text search. A schema registry with versioning. A cluster layer that places data on a consistent hash ring and replicates metadata through CRDT gossip. Three deployment topologies — standalone, hybrid, cluster — selected by configuration. The consistency model is specified in TLA+ and model-checked; a research whitepaper accompanies the documentation.
How to get it
Architecture, consistency model, formal specification and whitepaper are public at data.mlambda.net. Builds are available through early access; production use under commercial licence.