Alvaro Herrera <alvherre@xxxxxxxxxxxxxxxxx> writes: > Excerpts from Richard Huxton's message of mié sep 01 16:39:55 -0400 2010: >> OK - so not fillfactor and not some unicode-related padding. I can't see >> how a 32 vs 64-bit architecture change could produce anything like a >> doubling of database size. > Depending on table schemas, why not? e.g. consider a table with a > single bool column. It will waste 7 bytes on 8-byte MAXALIGN machine > but only 3 on a 4-byte MAXALIGN machine. Yeah, but after you account for row header overhead, the worst-case percentage bloat still should be a lot less than 2X. It would help if Devrim could break down the bloat to the level of individual tables/indexes. regards, tom lane -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
