On Tuesday 13 July 2010, dotnetdub wrote:
we're using a pool of identical kamailio load balancers in front of our systems, requests are distributed to it by routing means or DNS. If one of this system fails or needs to be removed out of the pool (e.g. for maintenance) we just move its IP adress to one of the other active or spare machines. For more planned maintenance you can of course also use DNS to shift load from one machine to another.
Sorry to butt in on this conversation. What is the recommended way to achieve failover and redundancy?
I know of some solutions running Heartbeat and DRBD and also some people use SRV records and DRBD or MYSQL replication.
What is the recommended way of achieving high availability SR scenarios.
Hi Brian,
well, i don't think there is one canonical solution to this problem. We use basically everything of the technologies you mentioned above in one part of our system or another. ;-) It depends e.g. on your setup, your load, your customer requirements, your internal organisation..
Cheers,
Henning