> On Tue, Jan 31, 2012 at 10:53 AM, Tom Lehr <
tomc...@gmail.com> wrote:
> > Hello All
>
> > Have been doing some reading about clustered indexing and wondering
> > what experiences other folks have had with them?
> > We currently do not use them in our high transaction OLTP world and am
> > a bit wary due to reading that they have to be "re-clustered after
> > table alters and are disturbed by ongoing updates" (from the manual,
> > although it does not specify much in details in regards to the ongoing
> > updates statement).
In my previous life, I supported a reservation system for a hotel
chain. The inventory and rates information was maintained in distinct
tables per property. Rates could be added in any order at any time so
they would be inserted haphazardly.
My goal was to reduce IO to these two tables because the vast majority
of lookups would be going against those two tables. Also, the
majority of the rate/inventory was against data in the next two weeks,
even though we had two years of future inventory and rates.
So I would periodically re-cluster those tables by date. That way I
could minimize the number of pages that had to be loaded into the
buffer for rates and inventory. And was able to have a high write as
well as read cache hit rate.
>
> > thanks in advance,
> > tom
> > _______________________________________________
> > Informix-list mailing list
> >
Informix-l...@iiug.org
> >
http://www.iiug.org/mailman/listinfo/informix-list