EXPLAIN deep-dives, JSONB and pgvector tricks, Postgres operations, and what we're shipping in ide99 — the desktop PostgreSQL IDE.https://ide99.ioen-us© 2026 ide99https://ide99.io/blog/reading-a-postgres-explain-planhttps://ide99.io/blog/reading-a-postgres-explain-planEXPLAIN (ANALYZE, BUFFERS) shows what Postgres actually did, not what it guessed. Read the plan from the most-indented node outward, find the node with the highest self-time, and compare rows estimated vs rows actual. A Seq Scan on a small table is fine; on millions of rows on the hot path it's usually a missing index.Thu, 18 Jun 2026 00:00:00 GMTPerformanceexplainindexingquery-planningperformanceexzvorhttps://ide99.io/blog/working-with-jsonb-in-postgreshttps://ide99.io/blog/working-with-jsonb-in-postgresJSONB stores parsed, binary JSON you can index and query in place. Use the path operators (->, ->>, #>>) to read, the containment operator (@>) with a GIN index to filter fast, and jsonb_set to update one key without rewriting the whole document. Prefer jsonb over json unless you need to preserve key order and whitespace.Sun, 14 Jun 2026 00:00:00 GMTJSONBjsonbindexingginquery-planningexzvorhttps://ide99.io/blog/autovacuum-explainedhttps://ide99.io/blog/autovacuum-explainedAutovacuum reclaims space from dead row versions left by MVCC and refreshes the statistics the planner relies on. It triggers per table when dead tuples cross a threshold (a base value plus a scale factor of the table size). On big, write-heavy tables the default scale factor is too lazy — lower it per table so vacuum runs more often, in smaller bites.Thu, 28 May 2026 00:00:00 GMTOperationsvacuumautovacuumbloatoperationsmvccexzvor