From: Mark Cave-Ayland Date: Thu, 14 Apr 2011 10:37:00 +0100 Subject: [postgis-users] (no subject) In-Reply-To: References: Message-ID: <4DA6C03C.1060600@siriusit.co.uk> On 12/04/11 17:38, Paul & Caroline Lewis wrote: > Hi, > Thank you Mark and Richard for your replies. Having looked at this it > seems a Full Vacuum is the answer, however I'm not sure why. Processing > the SQL scripts as originall reported I do get a large table from > TestSet1 and a small table from TestSet2. Once a Full vacuum is > performed on the large table from TestSet1 its size drops to the same as > the small table from TestS2, however adding a full vacuum into the > TestSet1 procedure makes it slower to run than TestSet2, very much > slower especially on uploading the very large data sets (70 mill rows). > This begs the question is TestSet2 very efficient or is it missing > something fundamental that a Full Vacuum provides that I'm not realising > at the moment. That's strange - do you see the same behaviour if you swap the order of the data load, i.e. do the ordered data set first, and/or use a different table name for each load? I'm just wondering if you're seeing some kind of database bloat if VACUUM fixes the issue. ATB, Mark. -- Mark Cave-Ayland - Senior Technical Architect PostgreSQL - PostGIS Sirius Corporation plc - control through freedom http://www.siriusit.co.uk t: +44 870 608 0063 Sirius Labs: http://www.siriusit.co.uk/labs