Oneconnect profile + SNAT + Persistance

Oneconnect is more effective with SNAT because there are more server-side connections that can be re-used. If you don't use SNAT, then we can only reuse backend connections that came from the same client-IP address.

The question of whether to use oneconnect with persistence is specific to your application. They don't help or hurt each other. If you need persistence, then use it. That decision is really separate from the oneconnect decision.

As far as packetflow, when a client connection comes in to a virtual that has oneconnect and SNAT, F5 makes a lot balancing decision based on the virtual server, pool, and iRule logic. Once that is done, it looks to see if it has a server-side connection open already and available to use. If it is available, it will use it. If not, it will open another one.

Here's is a detailed description: https://support.f5.com/kb/en-us/solutions/public/7000/200/sol7208.html