"Hilary Cotter" wrote:

> 1) Yes, an incremental population can take as long or even longer than a
> full population. You are best to use change tracking with update index in
> background if you are using SQL 2000. The reason for the approximate same
> length of time, is that for a full population the textual content in each
> row is extracted and indexed. For an incremental population each row is
> extracted and checked to see if the timestamp has changed. SQL FTI does this
> because it has to know which rows have changed, been deleted, or been
> inserted. For large amounts of changes/inserts/deletes you will find that
> incremental populations can take longer than full populations.
>
> 2) I would use change tracking with scheduled indexing. You will find
> performance to be much better than incremental populations.
>
> --
> Hilary Cotter
> Looking for a SQL Server replication book?
> http://www.nwsu.com/0974973602.html
>
>
> "Hugo Venancio" <HugoVenancio@discussions.microsoft.com> wrote in message
> news:C7C1630D-F315-48E2-B448-0BDA154279EA@microsoft.com...
> > Hello everyone,
> >
> > First of all, my name is Hugo.
> > I'm having a very strange situation when performing an Incremental
> Population.
> >
> > I have a Table "A" with a timestamp column, and a catalogue over three
> > columns of "A". This table has only 100K rows (a subset of our production
> > environment, with about 3 Million rows).
> > Just after performing a Full Population, I've started an Incremental
> > Population, and the time was exactly the same on both operations! I didn't
> > change any row (no updates, no inserts, no deletes, nothing was done), so
> the
> > timestamp didn't change as well.
> >
> > 1. Was it expected? I was expecting to see the Incr. Popul. taking less
> time
> > to finish than a Full one! What do you think I forgot?
> >
> > 2. I'd also appreciate your guidance on the following question:
> > This table has customers information and (once in production, with 3
> Million
> > rows) it's supposed to be updated every 48h on about 5K rows (50% updates,
> > 30% inserts, and 20% unfortunately deletes!!!).
> > What do you think it'd be the best approach?
> > Should I run an Incremental Population after that table update (every
> 48h),
> > or switch the Change Tracking on with scheduled Re-Indexing? After the
> above
> > strange behaviour, I'm affraid that an Incremental Popul. would take as
> long
> > as a Full Popul.
> >
> > Ah! I'm using SQL Server 2000 on a 2.8 GHz 4 processor box with 1GB of
> RAM.
> >
> > Thanks for your precious help in advance.
> >
> > BR,
> > Hugo
>
>