Posted by Jochem Maas on 10/28/05 11:14

Richard Davey wrote:
> Hello Jochem,
>
> Friday, April 22, 2005, 8:13:15 AM, you wrote:
>
> JM> in a round about way he seems to be asking 'why'? which is not
> JM> totally invalid.
>
> To be honest when I ask people "why?" I actually use that word (or at
> least something closely related to it).
>
> JM> do you want/need a master/slave configuration
> JM> or true decentralisation and/or two way syncing?
>
> See my reply to Drewcore for far more detail which answers the above.
>
> JM> sounds like a costly undertaking, not to be taken lightly... I'm
> JM> interested to understand the requirement that dictates such server
> JM> redundancy. Given the implied cost, planning/implementation should
> JM> probably left to a company who already knows the answers to your
> JM> questions :-/?
>
> This isn't about redundancy, it's about enhancing the experience for
> customers physically located thousands of miles away from the server
> they are trying to access. So, bring the content closer to them and
> drop their wait times massively. It's easy for those of us sat on the
> end of cable connections to become complacent about this IMHO (i.e.
> "the Internet is fast enough now that you don't need to do this"), but
> in reality that's not really yet the case.

interesting problem you have. the first thing that jumps into my head is
Squid... http://www.squid-cache.org/

Squid is really powerful and a bitch to get to grips with ;-).
If you were to take the 'Squid' route you would only be replicating (caching)
you main servers output rather than having to replicate DBs/source-files etc.

good luck which ever route you take.

>
> Best regards,
>
> Richard Davey

[Back to original message]