Re:Hstore development for 9.4 release
by sudhirj on Thursday November 07, 2013 @11:03AM
(#6690094)
When map/reduce comes into play, speed isn't usually the main issue. It's a way to do work in parallel - which may be done to increase speed, but more often than not it's done because there's too much work to be done any other way.
Put another way, I see map/reduce as a background task - building indexes, reporting, etc. I don't think speed is as important there as it is for a foreground task like GET, PUT and DELETE.
For these operations, especially by key, Postgres has always been right up there - what's happening now is that Postgres is getting really good at NoSQL features - documents, key value pairs, deep querying etc - good enough that if you have Postgres you can do pretty much whatever you could have done with Mongo - while still making use of everything else it has to offer.
Put another way, I see map/reduce as a background task - building indexes, reporting, etc. I don't think speed is as important there as it is for a foreground task like GET, PUT and DELETE.
For these operations, especially by key, Postgres has always been right up there - what's happening now is that Postgres is getting really good at NoSQL features - documents, key value pairs, deep querying etc - good enough that if you have Postgres you can do pretty much whatever you could have done with Mongo - while still making use of everything else it has to offer.