Fioccola Giuseppe <[email protected]> writes: > We have refreshed this draft on Alternate Marking application in NVO3 > and updated the reference to RFC8321.
I was reading somewhere (and it must have ultimately been linked from NVO3) of a scheme that provided the benefits of two-bit marking, but used only one bit. The concept was that in the middle of each block of packets with a single value of the L bit, a single packet has the bit inverted: 0 0 0 0 0 1 0 0 0 0 0 1 1 1 1 1 0 1 1 1 1 1 0 0 0 0 0 1 0 0 0 0 0 1 1 ... The idea being that examining the stream of packets, as long as the loss rate is low, the packets with the inverted bit can be identified because all of the nearby packets have the opposite L value, and the observed L bit values decomposed into the effective L and D bits: Lobs: 0 0 0 0 0 1 0 0 0 0 0 1 1 1 1 1 0 1 1 1 1 1 0 0 0 0 0 1 0 0 0 0 0 1 1 ... Leff: 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 ... Deff: 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 ... >From there, processing works as in two-bit marking. If this works in practice, it would be advantageous because it requires only one bit from each packet, and such bits are always in short supply. In the case of Geneve, there are 6 bits currently reserved, and it would be helpful if PM marking only consumed 16% of them. Dale _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
