Re: newbie on cluster

[Matthew Marlowe]

> 1) If I deploy MySQL cluster on the two boxes
> located in the primary location, do I ensure that
> the data is highly available and a load balancing
> will also happen between the two boxes?

You will want to provide some form of loadbalancing
yourself to fully utilize the cluster.

> 2) Can I use MySQL replication of this cluster
> with the 3rd box that is located in a different
> place so that I get almost all the data in case of
> a disaster in at the primary location?

My understanding from the sessions at the recent
user conference is that it is not possible to combine
standard replication with clusters at this point. The
async vs sync models just dont work well together
right now. There is a plan to unify the whole replication
architecture in a future release, but I'm not sure when
that will be.

> 3) I am planning for the cluster because, as per
> my understanding, if I replicate the databases
> between the boxes in the primary location, there
> might be some time lag between the two boxes being
> synchronised and hence in case of a failure of one
> of the boxes, the other one might be slightly out
> of sync. Am I correct?

Yes, typically with replication architecture, which is
async, you are always going to have issues with delays
so cluster is better. But, in your case - if you really
need to have multiple sites up and running across
long distances, cluster probably isn't going to work since
it doesnt work with replication and is extremely bandwidth
sensitive.