Rapid Recovery

Direct connect core to server(s)

Have any of you tired this? I though of setting up a practice DR destination server in the rack and connect it directly to the core with a dedicated switch or crossover cable to keep it off the LAN. It then it occurred to me I could actually connect all of my servers back to the core each on it's own NIC to keep backup overhead off the LAN.


  • This is fairly common in larger environments. They have a dedicated nic for backup traffic. Most backup products have a setting that allows you to enter in a preferred nic to send traffic over but rapid recovery does not. In order to force this, you would have to use the IP address (vs hostname) of the nic you want when you add it to protection. And then if that nic fails, backups fail
  • If I changed how the existing servers connected would it interrupt the backup chain?
  • No, you can remove the client but KEEP recovery points, then add the client back in using IP and the core should bring it back. There can be issues with this process and they have some TN that describe how to fix it
  • It's easier than that. Just go into the settings on the agent and change the hostname to the new IP. The core will use the new IP to connect to the agent. As long as the network is there to use and the core can connect to the agent on that IP, it will act like nothing changed. If the networking is set up wrong, the agent will go offline until the networking is fixed and functional.

    DO NOT REMOVE THE MACHINES FROM PROTECTION. That forces a new base image because it tells the protected machine to stop logging changes.
  • This would radically change a lot of things for the better. I schedule new snapshots and recoveries around user load because of heave network overhead when doing these things.

    I'll try it and report back for posterity sake if nothing else.