skaidb documentation
skaidb is a leaderless distributed SQL database: every node serves reads
and writes, replication is quorum-based with tunable consistency, and
clustered UPDATEs are linearizable per row. One binary gives you SQL,
secondary/global indexes, full-text search, vector search, geo queries,
and time-series storage.
Getting started
| Guide | What it covers |
|---|---|
| INSTALL.md | Installing from the apt/rpm repo, first start, systemd |
| DOCKER.md | Running in containers |
| QUERY_SYNTAX.md | The complete SQL dialect — DDL, DML, SELECT, transactions, SHOW |
| HOWDOI.md | "How do I…?" — task-by-task code examples in every client language |
Using the database
| Guide | What it covers |
|---|---|
| INDEXING.md | Secondary indexes, planning, covering counts |
| GLOBAL_INDEXES.md | Value-sharded (global) indexes on clusters |
| SEARCH.md | Full-text search: analyzers, hybrid ranking, highlights, suggestions |
| VECTOR.md | Vector/ANN search, quantization, managed embeddings, reranking |
| GEO.md | Geospatial indexing and queries |
| TIMESERIES.md | Time-series tables, rollups, PromQL/remote-write, Prometheus API |
| PROCEDURES.md | Stored procedures: CREATE PROCEDURE, CALL, privileges |
| MCP.md | MCP server: exposing the database to an LLM, and the security model for it |
Operating a cluster
| Guide | What it covers |
|---|---|
| CLUSTERING.md | Running a cluster — the overview and map of the pages below |
| CLUSTER_SETUP.md | Forming a cluster: members, ports, seeds, verifying the ring |
| CLUSTER_CONSISTENCY.md | Replication factor, consistency levels, linearizable UPDATEs, write errors and retries, ordering, clocks |
| CLUSTER_SMALL.md | One, two and three members: what each buys and what a member outage costs |
| CLUSTER_PLACEMENT.md | Per-table placement and pins, partition co-location, transactions, foreign keys |
| CLUSTER_MEMBERSHIP.md | Adding, decommissioning, forcing out and rebuilding members |
| CLUSTER_OPERATIONS.md | Rolling restarts, backups, point-in-time recovery, damage, memory, disk, shutdown, limits |
| CLUSTER_REPAIR.md | Anti-entropy, hinted handoff, REPAIR CLUSTER, RECLAIM, CHECK INDEX |
| CLUSTER_SECURITY.md | Internode auth and certificate rotation, client TLS, at-rest encryption, KEK rotation |
| CLUSTER_WITNESSES.md | Witness mirrors and BI nodes |
| RESHARDING.md | Joining/removing nodes, placement changes, data movement |
| METRICS.md | Every Prometheus metric, with alerting hints |
| GRAFANA.md | Dashboards via the scoped Prometheus API |
| UI.md | The built-in web UI |
| KERBEROS.md | GSSAPI/Kerberos authentication (server, clients, REST SSO) |
| STREAMS.md | Change streams: capture, replay and subscribe to matching writes |
| MQTT.md | Native MQTT 3.1.1/5.0 broker (IoT clients, replicated broker state, topic ACLs, topic→table sink) |
Reference
| Document | What it covers |
|---|---|
| LLM.md | The complete machine-readable reference — everything in one file (raw Markdown, made for LLM context windows) |
| GLOSSARY.md | Every abbreviation used in these docs, and what it means here |
| PROTOCOL.md | The binary wire protocol every official driver implements — framing, SCRAM handshake, opcodes, value encoding |
| BENCHMARKS.md | Measured comparisons vs PostgreSQL, MongoDB, MariaDB, Elasticsearch |