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

Reply via email to