[Serdev] usrloc loading

[Greger V. Teigre]

Hi guys!
I have taken the liberty to try to summarize your contributions, as well 
as add my 2 cents here:
http://www.iptel.org/failover_redundancy_and_scalability_overview

(this page will show up among others if you follow the Discussions link 
under Development for SER on http://iptel.org/ser)

Please feel to modify/add if I have misunderstood something or if there 
are arguments/descriptions missing.

Others: If you use or know of other scenarios, please add!!

I have written up the two scenarios described here and then added a list 
of variations (without describing which combinations go together). As we 
develop this further, we may want to split the two main approaches into 
several in order to be able to more specifically describe pros and cons etc.

Cheers,
Greger

Andreas Granig wrote:
> Andrei Pelinescu-Onciul wrote:
>> I prefer the first solution. Let's just say that we did not have
>>  a good experience with usrloc + db cluster from the performance point
>>  of view.
>
> It needed some parameter tweaks in the cluster configuration to 
> perform reasonable, but once optimized, it does its job very well, 
> especially in terms of performance and availability.
>
> For me the second solution seems to be easier to maintain and to 
> troubleshoot, because each node works on the same data, and no routing 
> between different proxy pairs has to be performed for on-net calls.
>
> But again, I think it's more a matter of taste than a matter of which 
> works better than the other.
>
> Andreas
>
>