https://www.postgresscripts.com/tags/io/Recent content in Io on Postgres ScriptsHugo -- gohugo.ioenPostgresScripts.comWed, 17 Jun 2026 00:00:00 +0000https://www.postgresscripts.com/post/query-postgresql-io-statistics-with-pg-stat-io/Wed, 17 Jun 2026 00:00:00 +0000https://www.postgresscripts.com/post/query-postgresql-io-statistics-with-pg-stat-io/ <h2 id="query-postgresql-io-statistics-with-pg_stat_io">Query PostgreSQL I/O Statistics with pg_stat_io</h2> <p>An I/O spike that nobody can attribute is a frustrating thing to chase. Was it autovacuum, a checkpoint, a bulk load, or ordinary client queries? Before PostgreSQL 16 the answer required stitching together several partial views; <code>pg_stat_io</code> gives a single cluster-wide breakdown of reads, writes, extends, and cache activity by backend type and I/O context.</p> <script async src="https://pagead2.googlesyndication.com/pagead/js/adsbygoogle.js?client=ca-pub-1012089347386563" crossorigin="anonymous"></script> <ins class="adsbygoogle" style="display:block" data-ad-client="ca-pub-1012089347386563" data-ad-slot="7945792173" data-ad-format="auto" data-full-width-responsive="true"></ins> <script> (adsbygoogle = window.adsbygoogle || []).push({}); </script> <h2 id="purpose-and-overview">Purpose and Overview</h2> <p><code>pg_stat_io</code> is a cumulative statistics view that decomposes the server's block I/O along two axes at once. The first is <code>backend_type</code> — client backends, autovacuum workers, the background writer, the checkpointer, and so on. The second is <code>context</code> — <code>normal</code> activity, <code>vacuum</code>, <code>bulkread</code>, and <code>bulkwrite</code>. The cross-product of those two axes is what makes the view powerful: it does not just say &quot;the server did 40 GB of reads,&quot; it says which subsystem did them and under what kind of operation.</p>