High Availability verses Recovery manager

[jonsilver]

I'm new to virtualization and have been reading and trying to understand the various products and capabilities, but I'm a little confused.

We have a VOIP switch that we want to run virtualized, but I'm not clear on the backup/recovery verses HA options. The switch does not go down (we run 7x24) so I'm not sure how we actually backup the system. I've read through the Recovery manager and it will back up to Crash consistent.

1. What do we need to do to make it Application consistent? Add hooks into the applications to flush their data to disk? The system has both Progress and Solid DBs. I think that one of them is memory resident. So adding hooks will be difficult. Are there any stanards?

2. If we run HA and the server goes down, from where does VMWare start the new VM Instance? from the VM file that is off on a SANs or from a backup?

3. I'm trying to understand the relationship between vCenter Recovery Manager and HA - do we use them together or are they indepenent?

thanks!

jon

[logiboy123]

SRM protects against datacentre failure.

HA protects against ESXi host failure by restarting the VM (from files on the SAN) automatically on a new ESXi host.

Fault Tolerance protects against VM failure by having a hot standby clone pre-provisioned on another ESXi host. If the first host goes down the second hosts VM automatically and seemlessly takes over. This option sounds like what you are after if the application cannot be clustered like an Exchange 2010 DAG can be. There are some caveats around this however; only 1vCPU can be used for instance.

Most products from VMware are compatible with each other but there are some exclusions, for instance a FT protected VM cannot also use HA as it is already being protected in a different way.

Regards,

Paul

[jonsilver]

I'm not sure what SRM is, but I'll try and search on it.

I understand the basics of HA and FT, but when would I used recovery manager? Is it only for going back in time to a specific recovery point?

If I have HA, then I'm going to my latest pre-crash point and so I guess that I would only need Recovery Manager to provide point in time recovery?

Is the above correct?  It also seems like Recovery Manager has some failure notification and manual restart capabilities to allow for an operator

to detect the failure and manually start a new instance - in which case, it's like HA but has a small down window?

thanks again,

jon