VMware Cloud Community
Bill_Oyler
Hot Shot
Hot Shot
Jump to solution

VRA questions

Hi folks,

I just have a few questions about the role of the VRA in vSphere Replication.  Assume for a moment a typical deployment with 2 vCenters, each with its own VRA.  Replication is all running from Site A to Site B (one-directional).  Assume replications are busy cranking through an Initial Full Sync from Site A to Site B.  Questions:

1) What happens to these jobs in the event that either VRA gets powered off?  What happens when the VRAs are then powered back on?

2) What happens to these jobs if the Source VMs being replicated are vMotion'd to a different ESXi host in the middle of their Full Sync?

3) What happens to these jobs if either VRA is vMotion'd to a different ESXi host?  (Specifically, the VRA at Site B, which I understand is receiving the data and sending to a specific ESXi host, which is then writing to the appropriate datastore?)

4) What happens if all but (1) of the ESXi hosts at Site B are put into Maintenance Mode and rebooted?  (Assume the one last standing ESXi host has full write access to all datastores containing VR replicas, and the VRA is running on the last standing ESXi host.)

My assumption in all cases is that the Initial Full Sync keeps running, doesn't skip a beat, doesn't start over, and just works, but I'd really like to know if this is or is not the case for the above scenarios.

Thanks,

Bill

Bill Oyler Systems Engineer
0 Kudos
1 Solution

Accepted Solutions
mvalkanov
VMware Employee
VMware Employee
Jump to solution

Hi,

I am assuming that VRA in your case is vSphere Replication Appliance, not the VR Agent at the source ESXi.

Replication traffic flows from the VR Agent (disk filter and others) at the source ESXi to the VR Server at the secondary site.

1) If the VR appliance (or an additional VR Server appliance) serving the replication at the secondary site is temporarily stopped, source hosts won't be able to send any deltas. When you power it back on, replication will continue from where it left off.

2) As already stated above - VR is compatible with vMotion and replication will continue normally.

There is an issue with Storage vMotion of a VM, if the source ESXi is 5.0 or 5.1 (fixed in ESXi 5.5) that might cause new full-sync to happen. Still checksum will be calculated and only the changed blocks will be exchanged.

3) vMotion of the VR appliance or additional VR server appliance does not affect ongoing replication.

4) VR server uses all available hosts to access a target datastore (mounted on them), if the currently used host fails for some reason.

Regards,

Martin

View solution in original post

0 Kudos
3 Replies
Biliana
VMware Employee
VMware Employee
Jump to solution

Hi,

Here are some answers to your questions, hope that you will find them useful.

1)  VRA is the component that resides on the source host. It is a module that is loaded on boot and it cannot be unloaded unless the host on which it is residing is powered off.

2) vSphere Replication is compatible with vMotion - replication should continue normally

3)I don't quite get this since the VRA is on the source site. It is sending the replication traffic to the destination VR server, which in turns applies on the target datastore. When configuring replication VR appliance instructs the source VR agent with the IP of the target VR server which is used as destination.

4) Replication will continue normally if there is at least one host that is able to access the target datastore.

mvalkanov
VMware Employee
VMware Employee
Jump to solution

Hi,

I am assuming that VRA in your case is vSphere Replication Appliance, not the VR Agent at the source ESXi.

Replication traffic flows from the VR Agent (disk filter and others) at the source ESXi to the VR Server at the secondary site.

1) If the VR appliance (or an additional VR Server appliance) serving the replication at the secondary site is temporarily stopped, source hosts won't be able to send any deltas. When you power it back on, replication will continue from where it left off.

2) As already stated above - VR is compatible with vMotion and replication will continue normally.

There is an issue with Storage vMotion of a VM, if the source ESXi is 5.0 or 5.1 (fixed in ESXi 5.5) that might cause new full-sync to happen. Still checksum will be calculated and only the changed blocks will be exchanged.

3) vMotion of the VR appliance or additional VR server appliance does not affect ongoing replication.

4) VR server uses all available hosts to access a target datastore (mounted on them), if the currently used host fails for some reason.

Regards,

Martin

0 Kudos
Bill_Oyler
Hot Shot
Hot Shot
Jump to solution

Thanks Martin, you are correct that I was referring to the vSphere Replication Appliance when I said "VRA".  This is all very good to know.  So to recap, it sounds like vMotion of the Source VM, or the destination VRA appliance, does not impact replication in any way.  Putting destination ESXi hosts into Maintenance Mode, so long as at least (1) host with access to datastores is up and running, won't impact replication in any way.  And shutting down the VRA appliance at the target site will cause replication to essentially "pause" until the VRA appliance is powered back on, correct?  In other words if it is in the middle of a 500 GB Full Sync, it will pause the full sync and then resume with the remaining 250 GB once the VRA appliance is powered back on?  It's also great to hear that Storage vMotion will not trigger a Full Sync in version 5.5.  Great information.

Bill

Bill Oyler Systems Engineer
0 Kudos