API Doc
Search Docs...
⌘ K
OverviewDeploymentManagementOperationReferenceGlossary

Monitoring data flows on a switch

The ACOS (AVE) cluster supports using NIC mirror mode based on port mirroring to monitor data flow on the switch, aiding in troubleshooting.

When creating or editing network devices for a virtual machine, you can enable Mirror mode for the virtual NIC. Once enabled, the port to which the virtual NIC is bound on the associated OVS bridge will always be configured as a mirror port.

The NIC continues to operate in mirror mode regardless of whether the virtual machine is shut down, started, or restarted. This ensures continuous monitoring of data flows connecting to the same virtual distributed switch on the host to which the virtual machine belongs.

Monitoring range

Once a virtual NIC has mirror mode enabled, it will only monitor virtual machine traffic going through the same virtual distributed switch connected to the host where this virtual NIC resides. It will not be able to monitor traffic from other virtual distributed switches connected to this host or traffic from other hosts.

To monitor traffic from virtual machines connected to different virtual distributed switches on the same host, you need to add multiple virtual NICs to the monitoring virtual machine. For each NIC, select a different VM network and enable Mirror mode to connect it to the corresponding virtual distributed switch and monitor traffic.

Restriction

  • Migrating virtual machines or rebuilding them on other hosts due to high availability may alter the monitoring scope.

  • Once Mirror mode is enabled for a virtual NIC, it will only receive mirrored traffic and will not be used for other purposes.

  • Once Mirror mode is enabled for a virtual NIC, network configuration via cloud-init is not supported.

Precaution

  • Once Mirror mode is enabled for a virtual NIC, it will be saved as an attribute of the virtual NIC in the snapshot, cloning, or template.

  • If a large amount of traffic is mirrored to a virtual machine being monitored, it may cause packets to be dropped. You can mitigate this issue by allocating additional CPU resources to the virtual machine.