Connection pools configuration¶
When deploying a connection pooler for the first time, it can be challenging to know how many connections to provision for each connection pool. The goal of using a connection pooler, like PgDog, is to reduce the total number of database connections by taking advantage of transaction mode.
Sizing up the right value for that setting depends on your database, but follows a few general principles.
One pool per user¶
PgDog creates one, independent, connection pool for each user entry in users.toml. If the pool_size setting is not specified, it will use default_pool_size instead.
This property is important to keep in mind, since adding several users will increase the maximum possible connection count on the database.
Estimating the pool size¶
To get the right setting for default_pool_size (or pool_size), take the current peak database connection count on your database and divide it by the number of users in users.toml. This will give PgDog the ability to open, at most, that number of connections, if needed.
Transaction mode connection re-use will bring that number down pretty quickly, but setting it that high to start with helps prevent any edge cases, e.g., slow queries, from starving the pool.
If you don't have reliable statistics for peak database usage, use the max_connections setting from postgresql.conf.
The formula¶
| Argument | Description |
|---|---|
default_pool_size |
Maximum number of connections PgDog can open to the database per user in users.toml. |
max_connections |
Maximum number of connections configured in the database. PgDog won't be able to open more connections than this setting. |
<number of users> |
Number of entries in users.toml. |