RFC1883 and ipv6 spec v2

[Deukyoon Kang]

At 03:53 PM 1/5/98 -0500, you wrote:
>Upon reading both Flow Label sections from RFC 1883 and
>draft-ietf-ipngwg-ipv6-spec-v2-01.txt I wonder about a few points.
>
>1) From my understanding the Flow Labels, once assigned by the originating
>node, will remain the same.  This represents a very interesting problem for
>backbone routers, 20-bit's is not very much address space and a backbone
router
>could foreseeably encounter two Flow Labels equal randomly generated by two
>different hosts.
>
>Upon discussing this with a friend.  We decided that if the label is
changed by
>the router after each hop, that the 20-bits for the label would be enough. 
>Providing (number of routers)^20th or even more if the Flow Label Table was
>done on a per interface basis.
>
>2) A rogue user could easily setup millions of flows to use up this space.
>
>3) Is the Flow Label supposed to be used in conjuction with the Hop-by-Hop
>routing options?
>
>Thanks, Darrell Newcomb
>
>

1) from my understanding, the flow labels shall remain the same from a
source node to a destination. yes, there's possiblity that two flows from
different hosts may be assigned the same flow labels. however, note that a
flow is identified uniquely by the combination of its source address and
flow label. 
	the idea of you and your friend was implemented in ATM switching. but it
requires either manual configuration of routers on the path from a source
to a destination or a signalling protocol. i guess that ipv6 desingers
hated both of ways. :)

2) yes. it's a possible scenario. but that kind of abnormal users could be
easily(?) tracked down if some security mechanism is enforced for the use
of the flow labels such as IP authentication header.

3) yes or not. it's up to you.

deukyoon.