bmanning at ISI.EDU
bmanning at ISI.EDU
Tue Aug 30 10:28:33 UTC 1994
A commentary on
"A Routing Design for ATM NAPS"
bmanning at isi.edu
23 August 1994
Attempts to clarify the important points in this document and proposes
an alternative that can be deployed in a reasonable timeframe.
This is an interesting and useful paper. It covers what was current
technology in 2q94 and points out one of several methods for
supporting ATM in the context of a Routing Registry/Route Server.
Several clever approaches were used to overcome gaps in standards
implementation by vendors.
The approach indicated in section two of the MERIT paper is a common
approach that was generally approved in both the NANOG meeting and a
NAP participant meeting that was held during the Toronto IETF.
The difficulties with the RFC 1490 approach were listed in the
Toronto meeting. I list them here:
There is no clear migration path from this approach to
a more correct/robust implementation using RFC 1577
encapsulation over native ATM (OC3c).
There are additional points of failure and additional costs
with the requirement for an ADSU.
The listed approach requires manual configuration.
This appears to lock the NAP into a top speed of 45Mbps
or DS3 speeds.
In addition to these weaknesses, there were a number of assumptions
regarding specific hardware implementations of ATM protocols (use
of a very small selection of vendors and limited interfaces) and
the ability and/or willingness of the attaching ISPs to work toward
a more elegant, cheaper, robust solution at the possible expense of
some time while vendors address the implementation gaps.
Questions on the desirability of a native media support with RFC 1577
support were posted to the IP-ATM wg of the IETF and the
email response was affirmative in recommending an approach with
RFC 1483/1577 over AAL5 with LLC/SNAP coding.
A brief review of the European PNO pilots seems to indicate that
this is the desired approach in that arena too.
Even the vendors who were not invited to either meeting have expressed
an interest in fixing these problems in the near term, where near term
is the next 2 to 6 months.
So, I suggest the MERIT paper should not be viewed as the only or
desired method of connecting to an ATM NAP. It should be carefully read
as the documentation of a successful experiment.
More information about the NANOG