Re: On-disk bitmap index patch - Mailing list pgsql-hackers

From Josh Berkus
Subject Re: On-disk bitmap index patch
Date
Msg-id 44C6652C.1020203@agliodbs.com
Whole thread Raw
In response to Re: On-disk bitmap index patch  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: On-disk bitmap index patch  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
Tom,
> (I'm also wondering whether this
> doesn't overlap the use-case for GIN.)

It does not.  GIN is strictly for multi-value fields.  I can think of 
applications where either GIN or Bitmaps would be an option, but for the 
majority, they wouldn't.

One particular compelling situation for on-disk bitmaps is for terabyte 
tables where a btree index would not fit into memory.   Index 
performance for an index which is 10x or more the size of RAM really 
sucks ... I can come up with some test results if you doubt that.

Also note that "low cardinality" is relative.  For a 1 billion row 
table, a column with 10,000 values is "low-cardinality", having around 
100,000 rows per value ... but that's still 0.01% of the table per 
value, making index use still applicable.

--Josh


pgsql-hackers by date:

Previous
From: Hannu Krosing
Date:
Subject: Re: On-disk bitmap index patch
Next
From: "Gevik Babakhani"
Date:
Subject: Re: root/administartor user check option.