What can I build?

Only combinations verified against the implementation. No marketing claims.

flowchart TB
    subgraph Models["one engine, three models"]
        R["relational tables"] --- J["JSONB documents"] --- T["time-series events"]
    end
    Models --> TX["transactions + MVCC"]
    TX --> A["order flows"]
    TX --> B["catalogs"]
    TX --> C["telemetry"]
    TX --> D["analytics"]
    D --> OPS["operated service\nprobes + logs + checkpoints"]

flowchart TB
    subgraph Models["one engine, three models"]
        R["relational tables"] --- J["JSONB documents"] --- T["time-series events"]
    end
    Models --> TX["transactions + MVCC"]
    TX --> A["order flows"]
    TX --> B["catalogs"]
    TX --> C["telemetry"]
    TX --> D["analytics"]
    D --> OPS["operated service\nprobes + logs + checkpoints"]

Transactional order flow#

CREATE TABLE + constraints + ON CONFLICT idempotent inserts + BEGIN…COMMIT multi-row writes + RETURNING echo + EXPLAIN ANALYZE validation. See Run transactions.

Document + relational catalog#

JSONB payloads + ->>/@>/JSON_TABLE queries + expression index on hot path + relational joins for ownership. See Query JSON.

Event / telemetry windows#

TIMESTAMPTZ events + ts index + now() - INTERVAL windows + date_trunc rollups + FILTER aggregates + keyset pagination. No continuous aggregates in v0.1.0 — roll up on read. See Query time.

Indexed analytics#

Selective WHERE + GROUPING SETS/ROLLUP + string_agg/array_agg/json_agg + ORDER BY … LIMIT. Verify with EXPLAIN (ANALYZE).

Operated service#

start.sh/Docker + db readiness/health/diagnostics probes + structured logs + checkpoint tuning + benchmark run regression checks. See Operations.