Performance implications of creating many, many sequences

From: Michael Gardner <gardnermj(at)gmail(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: Performance implications of creating many, many sequences
Date: 2010-10-22 19:18:34
Message-ID: FB36CB57-548C-4E35-A6EC-E44735C33B3C@gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Consider the following table:

CREATE TABLE invoice (
account_id integer NOT NULL REFERENCES account,
invoice_number integer NOT NULL,
UNIQUE (account_id,invoice_number));

I would like to do the equivalent of making invoice_number a serial type, but on a per-account basis. Would it be a reasonable approach to create a separate sequence for each individual account? Are there performance implications I should know about, given that there will be hundreds of thousands of accounts? Is there another approach I should be looking at instead?

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Rob Sargent 2010-10-22 19:50:26 Re: Performance implications of creating many, many sequences
Previous Message sunpeng 2010-10-22 18:50:58 Re: how to get the height of index tree?