Re: Fast insertion indexes: why no developments

Gavin Flower-2 wrote
> How about being able to mark indexes:
> 'MEMORY ONLY' to make them not go to disk
> and
> 'PERSISTENT | TRANSIENT' to mark if they should be recreated on
> machine bootup?

I would love that. But:

1) I'd like to make some tests with a "memory drive", and confirm that in
fact this would help (I'm sure I tried in the past, but I don't remember the
outcome)
2) I don't know if the fact that they are in memory should be handled by the
db or not. I was thinking about something more like "RECREATE IF NOT FOUND",
that is: if the files aren't found at postgresql startup, re-create the
index...
3) I don't know how many people would be interested (and how
doable/complicated that would be, considering log-replay, replication etc
etc)

--
View this message in context: http://postgresql.1045698.n5.nabble.com/Fast-insertion-indexes-why-no-developments-tp5776227p5776471.html
Sent from the PostgreSQL - hackers mailing list archive at Nabble.com.