URGENT : VMWare HA -Functionality Query

[psv]

Hello All,

I am new to to Virtualization Arena. We are evaluating VMware HA functionality for telco grade mission critical applications(availability and reliability requirements 99.999%).

I have following queries.

1). Is it possible to monitor particular applications/ process through VMWare HA. Are there any options to monitor selective processes.

2). Is it possible to configure number of local restarts for a process/service, before making a decision to move to another virtual machine.

3). Is it possible to have active-active/Hot-standby Configuration without restarting the VM & application as another VM. ie when the particular applications/process fails, we should be able to point to another VM, the application should continue from the last point of failure.

4). Does VMWare HA provides any checkpointing mechanism to preseve the in-memory data?

Appreciate your valuable inputs.

Regards,

Vijay

[FredPeterson]

VMWare HA is primarily a means to power back up a VM that has failed due to the host server crashing. The newer 3.5 features of monitoring a VM internally do not delve into the specifics you are looking for.

As for an Active / Active type of VM setup - that is a future function called Continuous Availability much like what Marathon offers - http://www.marathontechnologies.com/product_faqs.html, but completely outside of there being 3rd party software. If host A goes down - rather then restart VM A on Host B- VM B on Host B that was the Continuous Availability partner picks up almost immediately as if nothing happened. I have no idea on ETA for that.

[Yattong]

Sadly Vmware HA, High Availability is not what we have become accustomed to.

I had previously assumed that this is like a MS cluster type high availability. But it is not. VM state clearly that it is not a zero downtime piece of technology. ie: something must go down (an esx host in 3.0.2) for HA to kick into action.

In 3.5 there will be some sort of vm guest monitoring, so if there is a ms bsod, I believe that 3.5 is capable of restarting that vm.

Hope this helps

[jjkrueger]

Just to reinforce some of the points already made, VMware HA is definitely for High Availablility, not for Continuous Availability. in 3.0, HA would only come into play once there was a single host failure - the physical box had to die before any VMs were restarted anywhere. In 3.5, there is limited, experimental support for VM monitoring, but that functionality is more or less relegated to an occasional "Are you still running"-type check against VMware Tools in the guest. There still has to be downtime for that to kick in and restart the VM.

What if boils down to is that you're looking for guest-level continuous availability, and VMware really only currently has offerings for host- or VM-level high availability. VMware just isn't quite there yet.

The level of availability you're looking for may be found in on of the Marathon products listed in Fred's post, or if you're running ESX3 (not 3.5), Symantec does have Veritas Cluster Server available for ESX that may also meet some of your requirements.

-jk

[Kiran_M]

Using EMC's AutoStart, you can moniter a process and if that process is not responding AutoStart starts that process on another host.

This is like Application Level high availability and others are like Hardware Level high availabilty.

-Kiran