Re: [Detnet] I have some question for your draft.

["Pascal Thubert (pthubert)" <[e-mail]>]

Hello Yeoncheol

Thanks for raising the point.

I agree that there is a discrepancy but I think the detnet dataplane draft is correct and we need to work on the BIER-TE side.

Please see below :

As we discussed yesterday evening after the DetNet session, I am sending you my comment on the example shown in detent-dp-alt draft.


In page 32 of draft-dt-detnet-dp-alt-01, replication node A sends duplicated packets to nodes B and C. The packet going to node B has BitString = 00011110 and the packet going to node C has BitString = 01001110.

But, according to the BIER-TE forwarding process (draft-eckert-bier-te-arch-03, p12), node A must reset both p2 and p4 in BitString before sending. Then, nodes B and C will receive the same bitString 00001110.


I think you are referring to the text “



   Before

   sending any copy, the BFR resets all BitPositions in the BitString of

   the packet to which it can create a copy.  This is done to inhibit

   that packets can loop.


 “ and “





   BFR3 sees a BitString of p5,p7,p8,p10,p11,p12.  It is only interested
   in p1,p7,p8.  It creates a copy of the packet to BFER1 (due to p7)
   and one to BFR4 (due to p8).  It resets p7, p8 before sending.
“

DetNet needs BIER-TE to operate as described in the DetNet dataplane draft. A bit is reset when sending on the related adjacency. This is how the DetNet draft obtains traceability: the bits setting indicate which copy this is, by whom to whom, in a unique fashion. This is also how DetNet figures the transmission failures: if the bit arrives reset at destination, it means that this transmission worked over that adjacency. So with the DetNet operation, BFR3 resets p7 in the copy to BFER1 and resets p8 in the copy to BFR4.

OTOH I think that the DetNet operation fits the needs of BIER-TE to avoid loops. Once a packet has flown over an adjacency, it will not fly there again. So I think the change should be done on the BIER-TE draft. What do the authors think?

Pascal