The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2002-Jul> msg00039



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

AW: your mail

  • From: Eric Rosen <erosen@cisco.com>
  • Date: Tue, 02 Jul 2002 10:50:09 -0400
  • cc: Shahram Davari <Shahram_Davari@pmc-sierra.com>, "'Eric Osborne'" <eosborne@cisco.com>, George Sheng <george_s97@hotmail.com>, scullptor@yahoo.com, mpls@UU.NET
  • User-Agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.3(Unebigoryōmae) APEL/10.3 Emacs/21.2(sparc-sun-solaris2.6) MULE/5.0 (SAKAKI)


Heinrich> Independent from what  is clean and what is  quick&dirty, I cannot
Heinrich> detect any sense  in giving a hash-meaning to  the bottom label at
Heinrich> all.  

The only  purpose of the loadbalancing  hash is to reduce  the likelihood of
packets from the same end-to-end stream being sent on different paths, while
also not requiring  too many unrelated packets to be sent  on the same path.
Packets from  the same end-to-end stream  will tend to have  the same bottom
label,  and  at the  bottom  of  the stack  there  hasn't  been overly  much
aggregation.  So the bottom label seems  like the right thing to look at for
this  purpose.  In both  the L2VPN  and L3VPN  applications, this  works out
pretty well.

  • References:
    • AW: your mail
      • From: Hummel Heinrich <Heinrich.Hummel@icn.siemens.de>