Re: Free space mapping (was Re: Multi-Versions and Vacuum)

Alvaro Herrera <alvherre@atentus.com> writes:
> Tom Lane <tgl@sss.pgh.pa.us> escribi�:
>> Eventually we need code to collapse out free space in indexes.

> Maybe when some lower limit is reached in two consecutive pages they
> should be merged?

Yup.  The tricky part is to do this without causing problems for
concurrent readers of the index.

The Lehman-Yao paper that is the basis of our existing code explains how
readers can operate concurrently with splits of index pages.  It doesn't
address doing page merges concurrently with reader scans.  There are
newer papers in the literature that explain how to do that ... but no
one's gotten round to trying to implement it for Postgres.

            regards, tom lane