On Apr 5, 2011, at 17:08 , Thomas Narten wrote: > > What is *required* is that the hash function (or whatever function > that is used) on the router maps the tuples in a *uniform* way across > the range of possible outputs.
Then it seems like "Equidistributed Sequence" is the precise term you want. I too would like to see the case for Low Discrepancy vs. Equidistributed made more clear. One of the arguments I suppose could be made for Low Discrepancy is that it would provide a guarantee that *any* substring of the full 20-bit flow label could be taken as an input to a function that maps with low discrepancy to a particular load-sharing element. Equidistributed does not provide any such guarantee and requires A) that the domain of the function be the full label, and B) the function to map an equidistributed domain to a low discrepancy codomain. Are those functions any cheaper? I doubt it. -- james woodyatt <j...@apple.com> member of technical staff, core os networking -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
