I have numerous clients with VMware sites and we are working on a DR solution which we can bring accross the board. One issues that comes up always is the use of Sitee Recovery Manager. We are not able to impliment this even in PoC until we answer some basic fundemental questions.
From the time of early esx to current version 5.1 - if you take a snapshot of the VM (even without quiesceing but have "snapshot memory" options selected which is default selection) the VM will freeze for 3-5 seconds while that snapshot is processing, sometimes longer. Deleting snapshots i fount it not to be a problem only sometimes! This is also evident in backups causing VM Monitoring to reset VMs due to heart beat being offline for so long so we set it to custom settings.
I have read number of KB articles and there is nothing about this online only for NFS volumes and locking - this is iSCSI VMFS volumes so it doesn't apply. That being said - how would we impliment replication and site recovery manager when each time it snapshsots the VM it freezes it. Users would tear us a new ***** every 15min!
This is seen at all my clinets - they have no custom pre-freeze scripts or so on.
Use SAN-based replication or a software replication that doesn´t use snapshots with cbt.
Snapshots "kills" VMs that have high change rate on disks.
Good article describe the impact: http://www.vmdamentals.com/?p=332
The downside of this is that you don´t get crash-consistent data.