Re: MPLS OAM & the IETF

[Eric Rosen <erosen@cisco.com>]

Scott> My reading of the groups is  that one group, who are mostly concerned
Scott> with  the  transport of  IP  over  MPLS,  generally feel  that  tools
Scott> approximating the traditional "ping" and "traceroute" tools used on IP
Scott> networks are  sufficient.  A  second group seems  to feel  that those
Scott> tools do  not provide enough of a  view of the service  a customer is
Scott> getting  to be  sufficient and  that tools  approximating traditional
Scott> telephone system OAM tools are needed to get the complete picture. 

I would pick just two nits with this statement: 

- It's  not so  much that  the  first group  are concerned  mostly with  the
  transport of  IP over  MPLS, since  many in the  first group  are actively
  involved in the  PWE3 and L2VPN efforts.  It's that they  think of LSPs as
  being more like IP tunnels than like circuits. 

- The criteria  for something  falling  into the  category of  "traditional
  telephone  system OAM tools"  need to  be well  specified.  Every  time we
  propose  to work  on a  management tool,  we don't  want to  spend  a year
  debating whether it falls into that category or not. 

Assuming that we can specify  such criteria, the advantage of Scott's option
1 is that it excludes the  telco-style stuff from the IETF, thereby breaking
the logjams that we've been seeing in the this and other WGs.

While  option 2a  seems reasonable  at  first glance,  I don't  think it  is
implementable. 

In some ways, option 2b is  the most attractive; it doesn't prevent the IETF
from working on  anything, and doesn't constrain the IETF  in any way.  This
is in fact the status quo situation.  But it seems to be preventing progress
in the WGs.   Whether this is really a systemic problem,  or simply a matter
of personalities, is unclear.

So 2b  isn't working  out well, but  1 might  be overly constraining  to the
IETF. I  have to wonder  whether there is  something short of 1  which could
eliminate the WG logjams.  All we  really need to break the logjam right now
is a decision to pursue the ping and traceroute tools; maybe we don't need a
grandiose meta-chartering decision.