Don/Travis: thank you both very much for your responses.

My primary goals are these (in order of priority):

1. Stability of the system, overall (trying to get as close to 0 as possible the
chances of DB and filesystem getting out of sync).

2. Synchronization of the 2 consumption environments (as I would like to load
balance the 2 - any gotchas that you're aware of with doing that?)

3. Performance.



--- In intradoc_users@yahoogroups.com, "Glascock, Donald S."
<glascock.donald@...> wrote:
>
> I think Travis's observations are dead-on. Perhaps another perspective
> or restatement is: do you want the second consumption server to depend
> on the first consumption server for content that originated in the
> contribution server? While many might say "no", only you can make that
> call, since you know best the goals of your two consumption servers.
>
> Don
>
> ________________________________
>
> From: intradoc_users@yahoogroups.com
> [mailto:intradoc_users@yahoogroups.com] On Behalf Of T Biehn
> Sent: Tuesday, July 14, 2009 9:07 AM
> To: intradoc_users@yahoogroups.com
> Subject: Re: [intradoc_users] One-to-one-to-one, or one-to-many?
>
>
>
>
>
> It depends on loads and size of archiver jobs and the stresses on your
> Servers.
> It requires a lot of disk activity & some CPU to bundle them (DB Query
> to grab content, picking up each bit from the vault, packaging them in
> ZIP files)
> That being said, you should be absolutely fine with the first option,
> which will keep the consumption servers from being impacted (which i'm
> assuming is your goal) as long as you have both jobs set up to be
> automatic.
>
> If for some reason you find yourself doing any tables, please be
> extremely careful about replicating tables that you can't define
> timestamps on, as, on sufficiently large tables, this will ensure
> performance hit on both sides of the EQ.
>
> -Travis
>
> On Mon, Jul 13, 2009 at 7:08 PM, radychildrens<radychildrens@...
> <mailto:radychildrens%40yahoo.com> > wrote:
> >
> >
> > When using archiver to replicate content across multiple instances (in
> my
> > case, one contribution environment and 2 consumption-only
> environments),
> > which is better:
> >
> > 1. Push changes made to the contribution environment to both
> consumption
> > environments (2 separate archiver processes).
> >
> > OR
> >
> > 2. Push changes to the first consumption environment, then push
> changes from
> > the first consumption environment to the second?
> >
> > Or, is there really no difference in performance/stability/data
> integrity?
> >
> > Thanks!
> >
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>