High order bits, FWIW:

- this solution appears to ignore congestion-based losses
        large packets make that worse, not better

- this solution appears to ignore the impact on congestion control
        feedback systems react in chunks of an MTU, typically
        making that chunksize larger discretizes the congestion
        response, making it less smooth

        (this expands the point I said I'd explain later...)

- this solution should NOT interact with transport layers
        those layers can detect the interface MTU or changes
        to the extent they're prepared to already
        changes to the MTU will affect new connections already;
        affecting in-progress connections may not be appropriate,
        and is more complex than its corresponding benefit, IMO


Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to