Configuración y ajuste de Autovacuum

Welcome to tuning PostgreSQL's autovacuum! After learning about MVCC and VACUUM, let's dive into how to manage this crucial background process.

Autovacuum automatically reclaims space and updates statistics, preventing performance issues like table bloat and slow queries.

The autovacuum daemon runs in the background, constantly monitoring your database for tables that need attention. It performs two main operations:

• VACUUM: Reclaims space occupied by "dead" rows, making it available for new data.
• ANALYZE: Updates table statistics, helping the query planner choose the most efficient execution plans.

Most autovacuum settings are found in your PostgreSQL configuration file, usually named postgresql.conf. You can also change them at the database or table level.

Remember to restart PostgreSQL or reload the configuration for changes to take effect!

The most basic setting is autovacuum. It's usually enabled by default, and for most production systems, you should keep it that way!

Disabling it requires manual vacuuming, which can be easily missed, leading to severe performance problems.

Autovacuum triggers a VACUUM when a certain number of dead rows accumulate. This is controlled by two parameters:

• autovacuum_vacuum_scale_factor: A percentage of the table size (e.g., 0.2 for 20%).
• autovacuum_vacuum_threshold: A fixed minimum number of dead rows.

The vacuum triggers when (dead_rows > autovacuum_vacuum_threshold + table_rows * autovacuum_vacuum_scale_factor).

Let's say autovacuum_vacuum_threshold is 50 and autovacuum_vacuum_scale_factor is 0.2 (20%). For a table with 1000 rows, a vacuum will trigger when:

• dead_rows > 50 + (1000 * 0.2)
• dead_rows > 50 + 200
• dead_rows > 250

You can adjust these values for very active or very static tables.

Similar to vacuuming, autovacuum triggers an ANALYZE operation based on a threshold:

• autovacuum_analyze_scale_factor: A percentage of the table size.
• autovacuum_analyze_threshold: A fixed minimum number of changed rows.

Analyzing ensures the query planner has up-to-date statistics for optimal query plans, preventing slow queries.

Two more important parameters:

• autovacuum_naptime: How long autovacuum waits between checks on databases (e.g., '1min'). Shorter naptime means more frequent checks.
• autovacuum_max_workers: The maximum number of autovacuum processes that can run simultaneously across all databases. More workers mean more concurrent vacuuming/analyzing.

Sometimes, a table might need different autovacuum settings than the global defaults. For instance, a very large, frequently updated table could benefit from more aggressive vacuuming.

You can override most autovacuum parameters for individual tables using ALTER TABLE.

Great job! You've learned how to configure and tune PostgreSQL's autovacuum daemon.

• Autovacuum performs automatic VACUUM and ANALYZE.
• Key parameters control when (scale factor, threshold) and how often (naptime) it runs.
• You can customize settings globally in postgresql.conf or per-table using ALTER TABLE.

Proper autovacuum tuning is essential for maintaining database performance and preventing bloat.