IP provider performance measurement BOF
smd at sprint.net
Fri Mar 10 05:12:59 UTC 1995
In message <199503082144.QAA01126 at curtis.ansremote.com>, Curtis Villamizar writ
| IMO the IETF Operations Area has not been unfair or hostile toward
| providers. I'm not sure what the objections are.
I didn't mean to suggest that they are.
My objections are more philosophical and, perhaps, pragmatic.
I would like to see Internet standards-describing bodies
stay well away from trying to standardize operational
metrics and operational procedures.
This is to avoid three things:
-- operators who say, "our NIC is RFC 9000 compliant"
-- attempting to define what an operational
organization should do as an Internet standard:
"An ISP that is compliant with this standard
will..." which is particularly amusing if it
doesn't make business sense in a particular case,
or when it is drafted by people who mispredict
real operational issues (due to lack of foresight
or lack of experience) that obviate the standard.
This is essentially to avoid, "our competitor's
organization is not RFC 9000 compliant"
-- Operators and standards-folks are aliens from
another planet. Operators typically will bend
standards as they see fit for their business,
while standards-folks attempt to make it easy for
anyone -- say, someone new to the field -- to
interoperate with the current operators.
In extreme cases, operators completely ignore or
break standards willy-nilly, while standards-
folks get quite noisy when standards aren't
followed to the letter. I have run into both,
and, arguably, am an example of the former type.
There is a tension between the two mindsets which
is a good one for the Internet in general, but
which essentialy necessitates, in my opinion, a
division into two separate clubs when it comes to
working out issues that are more or less fully
in the domain of operators vs. standards folks.
That's not to say that standards-folks should
be discouraged from commenting on how we run our
parts of the Internet, nor that operators should
be told to shut up in working groups -- in fact,
I like to see operators in working groups and
standards-folks at NANOGs or IEPGs. However,
any kind of consesus-building about operations
issues or discussions about metrics showing good
or bad things about operators' networks should
be done in operators' forums, and not standards
Conversely, protocol standards shouldn't be done
at operators' forums.
An example of how things work differently among high
concentrations of operators vs high concentrations of
standards-folks is the CIDR issue. CIDRD at IETF and
the CIDR talks at NANOG are quite different, involve
different issues, and come up with very different forms
I believe this is a good thing, and should be encouraged.
So, wrt the provider measurement BOF, I am quite happy to
see the standards-folks describe standard tools which
operators can use to work out operational issues in
I would not like to see the Operations Area turn into an
olympics for providers any more than for vendors in other
More information about the NANOG