AD Replication

[Guest]

Archived from groups: microsoft.public.win2000.active_directory (More info?)

I have a customer that has a huge AD database (severl gig) and their
main site is where they do all their administration from. There
servers in the site where the admin work is done are constantly being
hammered with replication traffic. My theory is that because they are
using KCC intra-site repliction, which is change based, every time
someone makes a change to an object it is kicking off a replication
request.

I would appreciate others comments on this theory.

Thanks!

 

[Guest]

Archived from groups: microsoft.public.win2000.active_directory (More info?)

Intra site replication does not get initiated every time a change is made,
rather the DC that the change is made at stores them up for 5 minutes (or 15
seconds with 2003) before sending out a request to its replication partners
to pull the changes from it. Depending on how many admins and DCs you have
at the one site, it's possible they could all be using the same DC for
changes. You could try spreading the load a bit more by having some admins
perform their work on a different DC. The ADUC console can be targeted at
specific DCs for this purpose. It may help to draw out a diagram of the KCC
generated intra site topology and then determine at which DCs your admins
are making their changes, this may visually lead you to spot any obvious
bottlenecks. If you do find a problem, it's always possible to change the
topology manually although this should only be undertaken with great care
and forward planning.

"daeffinator" <marcy@daeffler.com> wrote in message
news:1118526188.009498.96260@o13g2000cwo.googlegroups.com...
>I have a customer that has a huge AD database (severl gig) and their
> main site is where they do all their administration from. There
> servers in the site where the admin work is done are constantly being
> hammered with replication traffic. My theory is that because they are
> using KCC intra-site repliction, which is change based, every time
> someone makes a change to an object it is kicking off a replication
> request.
>
> I would appreciate others comments on this theory.
>
> Thanks!
>