Summary:
Lower than expected network performance when using a small number of network connections.
Symptoms:
- Slow performance of network operations, such as backup over network, especially when using a small number of connections/channels.
- Slow performance of storage write operations, especially with a small number of disks per disk group.
- Slow performance of disk re-sync, especially with a small number of disks per disk group.
Affected products:
FlashGrid Cluster on Azure/AWS/GCP
Affected versions:
flashgrid-clan RPM versions below 20.10.147 (flashgrid-clan RPM versions 19.10.x and lower are not affected)
Root Cause:
Excessive number of network queues resulted in lower effective bandwidth available to each network connection.
Workaround:
If updating flashgrid-clan RPM immediately is not practical, then in some cases the performance can be improved by increasing the number of concurrent network connections. For example, for better backup speed over network, increase the number of RMAN channels.
Resolution:
Upgrade flashgrid-clan RPM to version 20.10.147 or newer on all nodes.
References:
To determine currently used version of the flashgrid-clan RPM, run rpm -qa | grep flashgrid-clan
The RPM update procedure must follow the recommended steps described in the knowledge base article: Updating FlashGrid software RPMs.