Scribd
Upload
17 views7 pages

Tuning Your PostgreSQL Server

400 connections is typically recommended as the maximum. Effective cache size is a guideline, not a hard limit. Shared buffers should be set to 1/4 of RAM, such as 8GB for 32GB RAM or 16GB for 64GB RAM. The number of connections increases with RAM amount. Work_mem and maintenance_work_mem control memory used for sorting and background processes. Checkpoint_segments controls how often data is written to disk.

Uploaded by

Stephen Efange
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
17 views7 pages

Tuning Your PostgreSQL Server

400 connections is typically recommended as the maximum. Effective cache size is a guideline, not a hard limit. Shared buffers should be set to 1/4 of RAM, such as 8GB for 32GB RAM or 16GB for 64GB RAM. The number of connections increases with RAM amount. Work_mem and maintenance_work_mem control memory used for sorting and background processes. Checkpoint_segments controls how often data is written to disk.

Uploaded by

Stephen Efange
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 7

400 for max connections is usually what is recommended max

https://pgtune.leopard.in.ua/#/
PostgreSQL cannot use a cache(shared buffer) bigger than 8GB as per rob and what he read in
PostgreSQL High Performance. ?

Effective cache size is just a guideline for the query planner.

Shared buffers should be a quarter of total ram e.g with 32gb Ram its 8GB and I tested with 64GB and it
was 16GB

The number of connections are originally 100 but as you move up the RAM in increases.
work_mem is for sorting/complex sorts. How much memory do you want to
allocate for complex sorting. This work_mem is alloted for any query
that comes in. so roughly work_mem = 55924kB that is 55mb per query.
So if multiplied by max connections it can eat up a lot of memory.

maintenance_work_mem is memory that will be used for most background


processes: pg_dump, pg_restore, vacumming, indexing and index
creation. So give this a good amount of memory or go with what pgtune
tell you.

Writing of data to disk is done in little segments and that’s where


checkpoint_segments come in. for every 3 segments there will be a
write to disk. Increasing this value will mean writs to disk will not
happen as often.
wal_buffers max value is 16MB used for writing transactions to disk.
Letting PostgreSQL Tell Us What's Wrong

http://www.craigkerstiens.com/2013/01/10/more-on-postgres-performance/

Read PostgreSQL 9 High Performance

https://www.youtube.com/watch?v=fva7qn9Yg3o

Indexes ?

Installing pg_stat_statements
The pg_stat_statements module provides a means for tracking execution statistics of all SQL
statements executed by a server.

pg_stat_statements is basically a monitor

https://www.postgresql.org/docs/9.4/static/pgstatstatements.html

https://pganalyze.com/docs/install/01_enabling_pg_stat_statements/
Restart the PostgreSQL server

You can reset statistics with:


https://github.com/jeffchulg/dbatoolbelt/tree/master/postgres

Top Queries

http://www.craigkerstiens.com/2013/01/10/more-on-postgres-performance/

SELECT
(total_time / 1000 / 60) as total_minutes,
(total_time/calls) as average_time,
query
FROM pg_stat_statements
ORDER BY 1 DESC
LIMIT 100;

Cache Usage

http://www.craigkerstiens.com/2012/10/01/understanding-postgres-performance/
SELECT
sum(heap_blks_read) as heap_read,
sum(heap_blks_hit) as heap_hit,
sum(heap_blks_hit) / (sum(heap_blks_hit) + sum(heap_blks_read)) as ratio
FROM
pg_statio_user_tables;
A cache hit ratio of 99% is good. Else if it’s lower increase your cache.

Index Usage

Fixing Slow Queries

You might also like