Sorting will be a lot faster on PostgreSQL 9.5: http://pgeoghegan.blogspot.com/2015/01/abbreviated-keys-exploiting-locality-to.html http://pgeoghegan.blogspot.

Sorting will be a lot faster on PostgreSQL 9.5: http://pgeoghegan.blogspot.com/2015/01/abbreviated-keys-exploiting-locality-to.html http://pgeoghegan.blogspot.com/2015/01/abbreviated-keys-expl... http://pgeoghegan.blogspot.com/2015/04/abbreviated-keys-for-numeric-to.html http://pgeoghegan.blogspot.com/2015/04/abbreviated-keys-for-... I wonder whether and to what extent Vitesse benefits from using this technique, which is fairly well known (e.g. it appears in a 1994 paper by Gray -- Alphasort). As I go into in the second link there, someone reported an order of magnitude increase in sort performance in one case. For what it's worth, I doubt that this fully accounts for why Vitesse is faster. I don't know that I'd trust my data to it, but I don't think it's snake oil, even if some of the figures shown are arguably a bit misleading.