db-failover-lab 1.0.0 — first measured snapshot

First measured snapshot: six engines, eight failover scenarios, run one cluster at a time on Docker (8 CPU, 12.5 GB).

• PostgreSQL 18 streaming replica promoted with pg_promote(); replay lag 383 µs (one same-host reading — a floor, not a cross-host latency).
• Redis 8 Sentinel: ~5 s to a writable replica, most of it the configured 5 s detection threshold.
• Redis 8 Cluster, MariaDB 11.8 + Galera 4, MongoDB 8.0 (replica set and hashed sharding), ScyllaDB 2026.1 and SolrCloud 9: what each preserved under node loss, in the README table.
• Not tested: network partition — one Docker host cannot produce a split network.

Full commands and output: RESULTS.md. Write-up: https://basantbhattarai.com.np/writing/failover-lab-six-engines-eight-scenarios/