4

From my understanding with failover clustering, only one machine is actually working, which means the other one(s) are sitting idle. So why would one use failover clustering instead of using a load balancing style with Windows Server 2012 to gain better performance with all the machines working as one yet still maintain high availability.

Do some situations work better with one over the other?

Jason
  • 3,931
  • 19
  • 66
  • 107
  • 3
    If your load balanced servers are each at 70% of their maximum capacity, and one fails, what happens? – Michael Hampton Apr 05 '15 at 02:29
  • 1
    @MichaelHampton Sometimes budget doesn't allow for sufficient redundancy. Then load balancing while both replicas are live could mean the system doesn't run slowly all the time, instead it only slows down while one replica is down. It could also be that there are more than two replicas in the pool. If you have five servers running at 60%, then what happens if two of them fails? Of course all of this doesn't take into account the situation where the load balancer fails, neither does it take into account that the service might be stateful in which case replication is anything but simple. – kasperd Apr 06 '15 at 17:09

1 Answers1

14

The world is full of applications that were never designed with clustering in mind and terrible things happen when you try to run multiple instances of the application concurrently.
But the application might be critical enough that having a standby server is a viable solution to meet your availability requirements.

Clustering software will typically ensure then that only one single instance of the application is running at any one time and will start the application on a backup node when the application on the primary node has stopped (either because of scheduled maintenance or because of unplanned events).

HBruijn
  • 77,029
  • 24
  • 135
  • 201