renghardt
commented
4 years ago Actually I think the following comment is also related:
- Gorry: prefer something slightly different. Service functions don't require a symmetric path - they just need to be able to observe both directions of the flow (not even necessarily observing in the same place - and possibly the forward/reverse path can differ after the function). Asking just to make sure authors are thinking about this.
Based on this, I would rephrase the following in our definition of "Service function":
"Some stateful service functions, such as NAT, require the same instance to be involved in both directions, i.e., on the path and the reverse path."
Currently it indeed sounds like the NAT requires the entire path to be symmetrical, which is not true, as Gorry points out.
In the June 2020 PANRG interim, Gorry commented that our current definitions of forward and reverse path are based on a routing view. There's other properties which can make a path "asymmetric", e.g., based on performance characteristics.
Maybe we should make it explicit that our definition refers to routing. Or on the same level of abstraction (as phrased by Spencer).