Memcached Replication

[Adam Lee]

Well, that's not really the way that memcached is intended to be used.
 You should think of it more as a cache that may or may not be there.

Having said that, we did end up abusing memcached in a way not
dissimilar to this.  If you have a usage pattern where writes are
uncommon, reads are the norm and you really want the data to be there,
you can hack a client such that sets always get written to every
server in a pool and reads are randomized throughout the pool.  It
serves to both load balance and replicate.

On Wed, Apr 2, 2008 at 1:31 PM, Kris Drebin <kris.drebin at a-cti.com> wrote:
> Good day everyone,
>
>  My development team has recently implemented memcached into one of our
>  high availability web application. It's working great! Now that it's
>  working properly, I would like to improve upon it's infrastructure. I
>  would like to have a secondary server at a completely different
>  service provider, for load balancing and failover purposes.
>
>  My question is:  Is there a way to mirror a memcached server such that
>  everything contained in one memcached server is contained in another,
>  in almost real time?
>
>  Thank you very much in advance! Your help is invaluable.
>
>  Regards,
>  Kris Drebin
>



-- 
awl