Tag: databases

ClickHouse vs DuckDB vs StarRocks: Light OLAP

ClickHouse, DuckDB, or StarRocks for home lab analytics? A practical guide to picking the right OLAP engine for your query patterns without losing the weekend.

Adding NOT NULL on a Big Table Without Downtime

How to safely add a NOT NULL column with a live backfill on a 200M-row Postgres table without locking your users out or scheduling any maintenance downtime.

Postgres HA: Patroni + etcd + HAProxy

Build a real Postgres 17 HA cluster with Patroni 4.x, etcd 3.5, and HAProxy — fully automatic failover, tested switchover procedure, zero manual babysitting.

Dragonfly vs Redis: Single-Binary Performance

DragonflyDB vs Redis 7.4 and Valkey 8.x: architecture deep dive, real performance benchmarks, honest gotchas, and when to swap your cache layer for a dragon.

KeyDB: A Redis Fork With Multi-Threading

KeyDB forked Redis and shipped multi-threaded I/O and active-active replication years before Redis itself did. Here is whether it is still worth using today.

pg_repack vs VACUUM FULL: Reclaim Bloat Live

Postgres table bloat killing your query performance? Compare VACUUM FULL vs pg_repack and learn how to reclaim disk space live without any scheduled downtime.

pg_stat_statements: What Most Tutorials Miss

pg_stat_statements is far more than a slow query log. Here is what most tutorials skip entirely: tail latency percentiles, cache hit ratios, and WAL bloat.

Postgres EXPLAIN ANALYZE Without Crying

Learn to read Postgres EXPLAIN ANALYZE output without losing your mind. Costs, row estimates, buffer hit stats, and the index and query fixes that matter most.

pgAdmin vs Adminer vs CloudBeaver: Web UIs

pgAdmin, Adminer, or CloudBeaver — three web UIs for your databases, zero overlap in who should use them. Pick the wrong one and you'll hate your life.

PostgreSQL on ZFS: Tuning, Snapshots, Pitfalls

Run PostgreSQL 18 on ZFS for atomic snapshots, lz4 compression, and silent corruption detection. Tuning guide, backup automation scripts, and the real pitfalls.

TimescaleDB Continuous Aggregates Without Pain

Stop melting your DB with Grafana queries. TimescaleDB continuous aggregates pre-roll your metrics so dashboards stay fast even at 90-day ranges without pain.

DuckDB for Local Analytics: Pandas on Steroids

DuckDB runs SQL directly on CSV, Parquet, and JSON files with no server and no import step required. Here is why it belongs in every home lab data toolkit.