Build the evidence before making the claim.
Orbita does not schedule 1.0 yet. The 1.0 claim is the product acceptance criteria holding with published evidence, and the project does not have enough measurements to date that responsibly.
Where the system stands
The KV surface, keyspaces, credentials and quotas, replicated WAL, epoch-fenced failover, replica reads, Admin API, CLI, deterministic simulator, object-backed partition format, multi-node registration, and Python end-to-end suite exist.
Manual partition split has end-to-end and simulation coverage. Automatic size-triggered splitting is not yet complete.
v0.1: make the architecture true
- Freeze and test the open partition format
- Complete resource visibility
- Run the control plane over a real Raft group
- Finish safe rolling-upgrade behavior
- Complete merge and managed voter work
v0.2: testing and operations
- Larger deterministic and real-cluster test campaigns
- Published correctness and performance evidence
- Backup, point-in-time restore, and an offline reader
- Dashboards, alert rules, capacity thresholds, and runbooks
- Standard gRPC health and server reflection
v0.3: coordination surface
- Watch and subscribe streams
- Tested locks, leader election, fencing, and epoch recipes
- Snapshot timestamps on reads
- Single-partition multi-key batches
v0.4: transactions and adoption
- Cross-partition snapshot isolation and read-set validation
- Official smart-client design
- etcd migration tooling and guidance
v0.5: security
- Client TLS and peer mTLS
- Prefix-level authorization
- Administrative audit log
- WAL and object-segment encryption
Later
Additional object stores, a Kubernetes operator, an experimental etcd API facade, multi-region work, and independent correctness analysis all follow the core production-readiness work.
Read the canonical ROADMAP.md for issue-level sequencing and current cut lines.