- Do not use basic auth while sending RestAPI requests to director. Use OAuth instead of basic auth.
- If we have a standby Versa Director available, and if the version is 22.1.4 or later, we can offload the live_commands to the standby director.
For example:
/vnms/dashboard/appliance/<Branch-Name>/live?
Sample:
/vnms/dashboard/appliance/Branch-Name/live?uuid=e03a0a62-d1ae-43a4-bda6-36dhe458d853&command=orgs%2Forg-services%2TEST-ORG%2Fsd-wan%2Fapplication-metrics%2Fbrief%3Fdeep%3Dtrue
- Do not call live-status API to get the status from all VOS devices via director. Instead, stream alarms and events from Analytics to your collector which will be sent in key-value format and consume that data to build your monitoring tool. Live API status may not give all the status changes that happened in between. To give an example, if interface flapped and came back in 2 seconds but APIs are made every 5 seconds once to get the interface status, interface flap alarm is missed.
- Limit monitor APIs to less than 50 concurrent APIs to the Director.
- If you do not follow Versa recommended sizing for directors, like not disabling hyperthread, not configuring 1:1 CPU pinning etc(refer Versa headend sizing document for recommendations), Director user-experience might suffer.
- Versa recommend to use streaming alarms and events from analytics to your collector and using that data instead of any pull models like API calls to Director or SNMP walk.
- Reach out to the Versa account and sales teams if you need any help.