Использование pg_stat_statements и pg_buffercache

PostgreSQL is incredibly powerful, and its functionality can be extended even further using extensions.

Extensions are modules that add new functions, data types, operators, and more to your database. They're like plugins that enhance your database's capabilities without modifying its core code.

In this lesson, we'll explore two crucial extensions for performance monitoring: pg_stat_statements and pg_buffercache.

pg_stat_statements is an invaluable tool for understanding your database's workload. It tracks execution statistics for all SQL statements executed by the server.

This extension helps you identify:

• Which queries are run most frequently.
• Which queries consume the most total time.
• Queries with high average execution times.
• Queries that read/write a lot of disk blocks.

It's your go-to for pinpointing performance bottlenecks at the query level.

To use pg_stat_statements, you first need to enable it in your PostgreSQL configuration.

1. Edit your postgresql.conf file and add pg_stat_statements to the shared_preload_libraries parameter. E.g., shared_preload_libraries = 'pg_stat_statements'.

2. Restart your PostgreSQL server for the change to take effect.

3. Finally, connect to your database and create the extension:

Once enabled, pg_stat_statements collects data in a view named pg_stat_statements. Here are some key columns you'll often check:

• query: The normalized SQL statement.
• calls: How many times the query was executed.
• total_time: Total time spent executing this query (in milliseconds).
• mean_time: Average execution time per call.
• rows: Total rows returned or affected.
• shared_blks_hit / shared_blks_read: Cache hits vs. disk reads.

Sorting by total_time helps find the biggest overall resource consumers.

Let's see how to find the top 5 queries that have consumed the most total execution time. This query helps you prioritize your optimization efforts.

The hit_percent column gives you an idea of how effective PostgreSQL's buffer cache is for that query.

While pg_stat_statements shows you query performance, pg_buffercache gives you insight into PostgreSQL's shared buffer cache. This is the memory area where PostgreSQL stores frequently accessed data blocks.

Understanding what's in the buffer cache helps you determine:

• Which tables or indexes are most actively used.
• If your shared_buffers setting is adequate.
• Whether queries are benefiting from cached data or hitting disk.

Enabling pg_buffercache is simpler than pg_stat_statements. It typically does not require modification to shared_preload_libraries or a server restart.

You just need to connect to your database and create the extension:

After creating the extension, you can query the pg_buffercache view to see which relations (tables or indexes) occupy the most buffers. Each buffer typically represents an 8KB data block.

This query shows the top 5 relations by the number of buffers they occupy in the cache:

If a table or index consistently appears at the top of the pg_buffercache output with many buffers, it means that data is frequently accessed and kept in memory.

This is generally a good sign, as memory access is much faster than disk I/O. A high hit_percent in pg_stat_statements often correlates with data being present in the buffer cache.

You can use this information to decide if increasing shared_buffers or optimizing queries to access less data would be beneficial.

Which of the following statements are true regarding pg_stat_statements and pg_buffercache?

Great job! You've learned how to leverage two powerful PostgreSQL extensions for performance monitoring:

• pg_stat_statements: Your essential tool for profiling SQL queries, identifying slow-running or frequently executed statements, and understanding their resource consumption.
• pg_buffercache: Gives you visibility into the shared buffer cache, helping you understand what data is actively residing in memory and how effectively your caching is working.

Combined, these extensions provide deep insights into your database's workload and memory utilization, guiding your optimization efforts.