• This topic has 3 replies, 1 voice, and was last updated November 14, 2022 by bfletcher@kirkland.com.

Failover RecoverTime much longer than FailoverTest

  • We have an 11 VM VPG that will become accessible relatively quickly when we run our failover test.  When we run an actual failover it takes about 90 minutes for all 11 VMs to come online.

    Can someone help me understand the different under hood on the recovery site when we run a  failover test vs. an actual failover? Specifically around how the Scratch, journal and vmdks behave.  I thought the only difference was that during the test failover it uses the sandbox network and does not give you the option to commit, whereas the real failover uses the prod network and gives you the option to commit or rollback.

    We have a case open to troubleshoot, I’m just looking for a high level explanation here.


    Does anyone have any insight with this?

    Hi Brian – during the pre-commit phase of a live failover, all things are essentially equal to a failover test (aside from the network settings). It is only once a user commits the operation during a live failover that things change. In other words, a pre-commit failover and rollback is essentially the same as a failover test that is started/stopped (aside from the network settings of course).

    I think this is covered in this video: https://vimeo.com/128476696 On quick review though, please do note that if WAN compression is enabled on a VPG, then the data written to the journal is also compressed; this enhancement was added after the original video was created.

You must be logged in to create new topics. Click here to login