Hi Uhmazing,
Here are a few tests I generally think about when I design an SD-WAN environment:
1) Are control connections up over all expected transports? ("Show control connections / Show sdwan control connections")
2) Are BFD (SD-WAN data plane) tunnels up to all expected remote SD-WAN routers? ("show bfd sessions / show sdwan bfd sessions")
3) Are all expected LAN routing protocol relationships formed at the site with the SD-WAN router(s)?
4) If above 3 conditions are met, perform critical application testing by using an end-user PC and navigating to the business critical applications.
5) Some customers choose to run iPerf or other speed capture / traffic generation tests before and after implementing SD-WAN at a site to compare results
6) Failover tests (if there is redundancy). Some examples include:
- Reload primary SD-WAN router at the site. Does traffic continue to flow? What was end user experience?
- Reload backup SD-WAN router at site. If this router is supporting TLOC-extension flows from primary router, does primary SD-WAN router reflect the failure and continue to pass traffic over directly-connected WAN transports
- If there is a single SD-WAN router and multiple transports, does failing a transport result in the router still being connected to controllers and still passing traffic over remaining TLOC(s)?
There are of course other failover tests, but these are obviously critical ones.
7) Ensure no traffic asymmetry is present if possible. The application inspection engine of the SD-WAN routers don't like asymmetry. Traffic should generally enter and leave the site via the same vEdge or cEdge router and use an active / standby setup, with the exception that the standby may be serving TLOC-extension to the primary.
Apart from this I can't think of anything major. The point is to make sure the box forms all expected connections, that the end user experience and critical applications work, and that failover works (if available).
Hope it helps!
Steve
