can you clarify "There is also no guarantee that the vm’s will not suffer any data damage during the process…."
As soon as MirrorView/a meet RPO and replicate the LUN and the VM inside, these VM should be in consistent state: is this correct or replicating LUN with VM inside could bring to an inconsistent state ?
What's the difference with RecoverPoint+AppSynch approach (except for all the features like DVR introduced by RecoverPoint) ?
MirrorView replicates the data that reside on the LUNs. It is crash-consistent.
What AppSync gives you is application consistency and not crash-consistency. It flushed the host to guarantee that the data is 100% reliable at the DR site. Without flushing from memory to disk the image is crash consistent.
GearoidG
251 Posts
0
October 4th, 2012 00:00
You have been busy!
(sent from my blackberry - please excuse any typos)
sullybags
45 Posts
0
October 10th, 2012 00:00
part 4 added...
Christian_Zippe
5 Posts
0
January 17th, 2013 05:00
Thank you for that great work. It really helps to understand this complex theme.
But for me there is one question open.
If i use Mirrorview/A how the virtual machine knows that it would been snapshoted in the array?
If the machine is not notified via vmware tools and/or a vmware snapshot i would have an inconsistent file system or database.
For a planned migration there is no problem because the vm would powered of and then the mirror would sync the delta.
But in case of unplanned failover it need the vm to start in an operational state for a given recoverypoint.
Does Mirrorview/A notify the VM may via VSI or do i need RecoverPoint?
Thx for your help.
sullybags
45 Posts
1
January 17th, 2013 06:00
In the case of a unplanned disaster recovery, VMware tools will *TRY* quiesce the file system of the vm before it shuts it down.
All vm’s will be shut down & powered back on in any SRM scenario.
All SRM will do is send command to the vm (to VMware tools) to do certain operations, it is not aware of the replication protocol you are using.
This is why It has to be tested ,in depth, before rolling out to production.
There is also no guarantee that the vm’s will not suffer any data damage during the process….
To be honest, I’m not familiar with recover point ,my lab was designed around was mirrorview…
gidiesse
2 Posts
0
July 30th, 2013 16:00
Dave,
can you clarify "There is also no guarantee that the vm’s will not suffer any data damage during the process…."
As soon as MirrorView/a meet RPO and replicate the LUN and the VM inside, these VM should be in consistent state: is this correct or replicating LUN with VM inside could bring to an inconsistent state ?
What's the difference with RecoverPoint+AppSynch approach (except for all the features like DVR introduced by RecoverPoint) ?
etaljic81
1K Posts
0
July 30th, 2013 19:00
MirrorView replicates the data that reside on the LUNs. It is crash-consistent.
What AppSync gives you is application consistency and not crash-consistency. It flushed the host to guarantee that the data is 100% reliable at the DR site. Without flushing from memory to disk the image is crash consistent.