and I will try to use the fourth too.
On 28 Ago, 13:56, Joe Webb <j...@webbtechsolutions.com> wrote:
> Hi Mauro -
>
> A couple of thoughts along these lines:
>
> 1) If you control the table that is the source of the events, you can add
> a column to it that indicates the last time the row was modified (or
> inserted). Then you can create a chronicles table that keeps track of the
> last time the generator ran and look for only those rows modified since
> the last time the generator ran. This works nicely.
>
> 2) A common way to check for changes to a row is to calculate the
> BINARY_CHECKSUM value across all the columns of the row and compare to the
> previous calculation (which would be stored in a chronicles table). It's
> not 100% because theoretically a row could be changed in such a way that
> it's checksum doesn't change, but that's highly unlikely.
>
> 3) You can use the SSNS stored procedures to insert events into the events
> table.
>
> 4) Or you can do it the way you have suggested.
>
> The best implementation will likely depend on your unique situation, but I
> would lean towards 1 or 2 if possible.
>
> HTH...
>
> --
> Joe Webb
> SQL Server MVP
http://weblogs.sqlteam.com/joew/ >
> On Tue, 28 Aug 2007 03:40:18 -0700, Peruukki wrote:
> > Hi all,
> > I am wondering what's the best practice for the problem I'm facing.
> > I'd like to use a scheduled provider that checks if a field is changed
> > in a table, but looking at examples it seems that there's no other way
> > but keeping a chronicle table with previous values.
> > Due to the large amount of data, it does not seem so performant.
> > So I was thinking at the possibility to trigger each change in sql
> > server, save data in a custom table and let the provider read this
> > user-created chronicle table.
>
> > Does it seem reasonable for you or is there a better way to do it?
>
> > Thanks Mauro