Re: CREATE INDEX uses INDEX ?

2015-03-08 10:04 GMT+01:00 Nicolas Paris <niparisco(at)gmail(dot)com>:

> Thanks.
>
> Then,
> Is it a good idea to run multiple instance of "create index on tableX" on
> the same time ? Or is it better to do it sequentially ?
> In other words : Can one seq scan on a table benefit to multiple create
> index stmt on that table ?
>
>
It usually is a good idea to parallelize index creation. That's one of the
good things that pg_restore does since the 8.4 release.

Nicolas PARIS
>
> 2015-03-07 12:56 GMT+01:00 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
>
>> Le 7 mars 2015 11:32, "Nicolas Paris" <niparisco(at)gmail(dot)com> a écrit :
>> >
>> > Hello,
>> >
>> > I wonder if the process of index creation can benefit from other
>> indexes.
>> >
>>
>> It cannot.
>>
>> > EG: Creating a partial index with predicat based on a boolean column,
>> will use an hypothetic index on that boolean column or always use a seq
>> scan on all rows ?
>> >
>>
>> Nope, it always does a seqscan.
>>
>> > Goal is to create partial indexes faster.
>> >
>> > Explain command does not work with Create index.
>> >
>>
>> You cannot use EXPLAIN on most DDL commands.
>>
>
>

--
Guillaume.
http://blog.guillaume.lelarge.info
http://www.dalibo.com