On Sat, 2005-04-23 at 15:04 -0700, Ron Mayer wrote:
> Bruce Momjian wrote:
> > See this TODO: * Allow data to be pulled directly from indexes
> > I think this is the direction we should be heading because it has more
> > general usefulness.
>
> I think read-only tables would have a few different types
> of general usefulness in addition to enabling index scans.
>
> Is this a fair summary of the potential benefits of READ-ONLY
> tables? (from both this thread and the archives):
>
> 1. Index-only scans are made possible fairly easily because
> you wouldn't need to check the heap for visibility.
>
> 2. Simple tables can be much smaller since you don't need
> most of the HeapTupleHeaderData. This reduction in
> space translates to a reduction in I/O through better
> use of the shared memory and OS caches.
>
> 3. A Read-Only Clustered table could be assumed to be
> sorted, so you could avoid some Sort steps
> for things like GroupAggregates and Merge Joins.
4. No need to VACUUM tables as part of a database-wide VACUUM, which is
particularly important for larger databases.
Best Regards, Simon Riggs