Thank you for replying!

The idea of sets (duh, yeah, of course) is a good one.  I didn't clue in on that.

How exactly to apply that idea with two nodes (total) where node 2 is essentially only in place as a 
log-ship-creating slave?  Still have the problem with all-inclusive-content log files as output, right?

How can I get the correct subscribed content of my logs to my users?  Node 2 is outputting log files 
every minute (presumably for oh, 50 sets, in your model, with one cluster).  How to go about 
separating those logs in a way that user A gets 25 sets and user B gets a different 10 sets and user C 
gets some intersection of what A & B are getting?

See what I mean?

Thanks again,
Jennifer


Andrew Sullivan wrote:
> On Wed, Aug 27, 2008 at 07:46:06PM -0700, Jennifer Spencer wrote:
>> I need help figuring out how to create a smart cluster layout to log-ship my data.
>> Please, oh please, help me understand a couple of things about nodes & 
>> clusters.  What is possible and what is not, specifically.
>>
> 
> I think you're missing a concept you need, which is "sets".
> 
>> 1.  The boss wants minimal drag on the production database (e.g. minimal 
>> triggers, minimal number of overlapping clusters).
> 
> Additional sets should not produce detectable drag.
> 
>> 2.  We have about 10 future subscriber sites that want sometimes the same 
>> data tables, sometimes different data tables from the same database (arg!).
> 
> Put the minimal set of tables that could be wanted at any one site
> into one set.  That site subscribes to that set, and not the others.
> 
> A
> 
>