Latency#

Transaction processing#

Transactions should not be run by nodes over a high latency network, if possible.

  1. Log in on any unified node (multinode unified topology) / application node (modular cluster topology).

  2. Run the following for all the other nodes:

    diag ping <IP>

  3. If any of the average round trip time is higher than 10ms, it is recommended that the nodes in the secondary DC should have voss workers set to zero (0) on version 11.5.3 and later, or the command app stop voss-queue is run on older versions.

Note

For software versions pre 11.5.3, the app stop voss-queue command needs to be run after every service restart or server reboot.

UC apps latency#

Latency between unified nodes and UC apps impacts overall provisioning times against the UC apps. This check is done as part of diagnosis of provisioning performance problems.

For each data center:

  1. Log in on any unified node (multinode unified topology) / application node (modular cluster topology).

  2. Run the following command for all the UC apps under investigation:

    diag ping <IP>

  3. Record the output, then send the file with VOSS L2 support.