The main concern I try to adress is when we have a large number of
heuristics trying to match on a large mumber of packets but making no match.

As an example one of the dissectors switched to heuristics default off
deals with communication between a controler and a drone. Which must be a
very rare case.

In the gui it is easy to turn all heuristics on or off. Making a selection
is more difficult as you would have to understand in what kind of
environment the protocols may be used.




Den fre 21 nov. 2025 09:49Guy Harris <[email protected]> skrev:

> On Nov 20, 2025, at 11:08 PM, Guy Harris <[email protected]> wrote:
>
> > Do we have any numbers on how much of a performance improvement results
> form disabling all heuristics?
>
> ...bearing in mind that disabling those heuristics could speed up
> dissection *because packets aren't being dissected past a certain point*.
>
> E.g., testing with a large NFS capture (NFS is recognized by its ONC RPC
> program number, not by being on port 2049, and ONC RPC is recognized by
> heuristics) would probably show a speedup because neither the ONC RPC
> dissector nor the NFS dissector are called, regardless of time spent with
> heuristics that fail.
>
> Note, though, that the ONC RPC dissector sets the "conversation dissector"
> for the TCP connection or UDP "connection" to be the ONC RPC dissector once
> it recognizes an ONC RPC packet, so that dissection of subsequent packets
> shouldn't involve the heuristics.
>
> _______________________________________________
> Wireshark-dev mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>
_______________________________________________
Wireshark-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to