Feature Request: high availability for EM

[mikeely]

As my experimentation has shown, it's impossible to have multiple running EM servers - only one EM can manage any given VBR. This means that right now in a distributed environment, Enterprise Manager is a SPOF. This absolutely should not be the case. Ideally a customer should be able to configure any number of EM servers, set up a round-robin DNS entry for them, and have fully available EM access. While the obvious use case for this is in a DR scenario - what happens when the site hosting the EM server gets hit by a beer truck? - the much more common use case for this is that admins should be able to access the EM console even if one of the EM instances is offline for some sort of maintenance. Operations shouldn't get in the way of access.

Obviously this will require some form of coordination and synchronization between EM instances, but I'd bet good money that the problem wouldn't be that hard to solve, especially with the VSA - you've got all the building blocks you need: set pgsql up in a hot-standby configuration, have some form of heartbeat configured to determine whether the primary server has gone out, and Bob's your uncle.

The goal in all systems is to reduce single points of failure to the degree possible, and this is one case that is solvable and needs to be done.

[RyanJ]

I doubt it will be added because EM will more then likely go away once the web client is full featured and everything EM can do, it will do. Won't be for a while, but its made EM development pretty stagnant.

[mikeely]

I hope that's not the case - the role EM provides is a single pane of glass for all Veeam backups within an org or business segment. Getting the B&R web client feature-complete to the Windows .NET client won't replace that functionality.