On Wed, Jan 30, 2019 at 6:17 AM Stewart Bryant <stewart.bry...@gmail.com> wrote: > > > > That's true for IPv4,the only way to do stateless ECMP and have > > fragments follow the same path as non-fragments is to hash over the IP > > addresses only. > > There is not enough entropy in that. > > I remember the original ECMP studies, and if the designers could have > got away with just SA/DA/Prot they would have. In those days we were > using s/w forwarders and the ECMP "feature" cost headline PPS which in > those days was the key metric. > > > > For IPv6 we can do better. The flow label allows finer > > grained per-flow routing, but still only requires inspection of IP > > header so keeping fragments in order just works. > > If you can trust it. > Stewart,
I trust it more than I trust that vendors will ever go out of their parse any transport layer protocols other than TCP or UDP to extract port information, or that they'll happily skip over extension headers or parse N levels of encapsulation to find anthe embedded transport layer, or that they'll figure out a reasonable way to deal with protocols that don't have enough information in plain text to discern a five tuple. Tom > - Stewart > _______________________________________________ Int-area mailing list Int-area@ietf.org https://www.ietf.org/mailman/listinfo/int-area
