Reusable distributed-systems algorithms without requiring a network framework.
It provides membership hashing, causality tracking, failure suspicion, and quorum aggregation as pure in-process building blocks.
Picks stable nodes for keys with:
- node weights
- optional bounded-load behavior
- deterministic hashing
- minimal remapping when membership changes
Tracks causal history across replicas:
- increment, merge, dominance comparison
- concurrent-version detection
- deterministic serialization
- configurable growth limits and compaction trade-offs
Computes a continuous suspicion score from heartbeat intervals:
- sample window and warm-up behavior
- injectable clock abstraction
- outlier handling
- deterministic simulations for tests
Suspicion is not proof of failure.
Collects concurrent replica results under a quorum policy:
- agreement and conflict detection
- early completion
- failure evidence
- cancellation of unnecessary work
- deterministic winner selection only when the policy allows it
- Client-side shard / replica selection
- Conflict detection for multi-writer records
- Heartbeat-based soft failure detection
- Reading from N replicas and waiting for quorum
- Shipping a cluster membership protocol or RPC stack
- Treating phi scores as hard failovers without host policy
dotnet add package FunctionFoundry.DistributedNone beyond the .NET shared framework.