Network congestion seen while vsphere replication performs delta sync operation
Article ID: 404557
Updated On:
Products
Issue/Introduction
You see a high spike in network congestion when VM's replicated by vSphere replication are performing delta sync operations on a guest OS that has Trim/Unmap enabled.
Environment
vSphere Replication 8.x / 9.x
Cause
When a guest OS VM issues unmap commands against a disk on the VM, all of the changed blocks that have been deleted will be sent to the target site as zero'd out blocks. This can cause network congestion during the sync operation as all blocks are being sent to the destination at once.
Resolution
Enable network compression on the replication. This will compress the data that is transferred over the network, packaging up the deltas to send as a bigger file instead of lots of smaller packets.
- Select the VM replication and select Reconfigure
- On the Replication Settings page select "Enable network compression for VR data"
- On the Ready to complete page, review all settings and select Finish.
The next RPO cycle for the VM will compress the data on the source ESXI host and then get sent to the destination.
Additional Information
Feedback
