Settings
Page Costs
http://archives.postgresql.org/message-id/BD2648C0-836B-4F4C-834A-4B26CF092B20@renci.org
http://archives.postgresql.org/message-id/07F01B2A-648A-4BA7-B056-EC4EA2FFE0D0@gmail.com
http://www.postgresql.org/message-id/CAPpwVW36gmHUG1d6KJ6tNyKXT7udJBsUDyUNf6gkBM_28rSe3w@mail.gmail.com [maybe? plan with slightly lower cost is 3x slower, so costing is off somewhere]
http://www.postgresql.org/message-id/3469.1367421052@sss.pgh.pa.us
http://www.postgresql.org/message-id/20121213152624.80080@gmx.com
Missing index
http://archives.postgresql.org/message-id/29599.1320244010@sss.pgh.pa.us
http://www.postgresql.org/message-id/511D0430.1090603@normation.com
Cost for @@ doesn't reflect it's true cost
http://archives.postgresql.org/message-id/4289.1308635599@sss.pgh.pa.us
work_mem too low
http://archives.postgresql.org/message-id/CAEGYRW7ecG6muQagLd=v_tCzpa148gdrKNbP6wNp_ER-6NZMAg@mail.gmail.com [TID bitmap lossifies]
Statistics Target Too Low
http://archives.postgresql.org/message-id/20459.1319337119@sss.pgh.pa.us [maybe]
http://archives.postgresql.org/message-id/4FD2B025.8080308@gmail.com
Statistics Target Too High
http://www.postgresql.org/message-id/515D92BF.5000309@axege.com
n_distinct estimates aren't accurate on large tables
vm.zone_reclaim_mode = 1
http://archives.postgresql.org/CAOR=d=3Pr9vYu7RC-7OyiOXVOTvvOJkdRo3_ahEUkmNF94YeZA@mail.gmail.com
out of date statistics cause bad estimates
http://www.postgresql.org/message-id/5114F15B.8060700@northplains.com
TOAST decompression is slow