Re: 802.1s mib

Just to be clear on the basics (which I think I am), I think you need to
mind your p's and q's or in this case s's and Q's.

IEEE Std 802.1Q tm, 2003 Edition incorporates IEEE Std 802.1s tm, -2002 so
.1s is history.  This is not to say it won't appear in many product
websites for years to come, even in places yet more inaccurately as .1S;
but strictly, it has gone.

And of course this is a difference between IETF, where RFC1157 refers to
the same string of ASCII in perpetuity and 802.1Q does not; the latter
needs a qualifying date.  So any .1Q MIB module needs a qualifier after it.

And while these IEEE publications do contain PICS proforma, they do not -
for me - contain much by way of conformance, certainly not the
macro-specified conformance that is now part of any IETF MIB module.  So
while the IEEE may have a good data model ready to be turned into a MIB
module, I think there is much work to be done in producing meaningful
subsets for remote management and IMO the actual IEEE document structure is
not a good place to start, ie MIB modules should cut across document
boundaries.  It then becomes unrealistic to talk of a .1s or .1Q MIB
module.

Tom Petch

-----Original Message-----
From: Harrington, David <[email protected]>
To: Tony Jeffree <[email protected]>; Congdon, Paul T (ProCurve)
<[email protected]>
Cc: Romascanu, Dan (Dan) <[email protected]>; [email protected]
<[email protected]>
Date: 03 April 2004 14:56
Subject: [Bridge-mib] 802.1s mib


>
> >On a related note, shouldn't the .1s MIB be part of an
> updated Q-BRIDGE
> >MIB instead of a separate MIB?
>
> Absolutely. S is now part of Q anyway.
>

Will all implementors of Q also implement S, or is it an optional part
of Q? If it is an optional part of Q, then it would probably be best to
write a separate mib that supplements the Q mib. This approach is
generally easier for implementors and users.

Note that you can have multiple mib modules defined in the same
document.

Updating mibs:

SNMP has some strict rules about versioning of mibs, detailed in the
SMIv2 documents (RFCs 2578, 79, and 80). It is important to follow these
rules to ensure interoperability between agent implementations and
manager implementations, which are often implemented by different
vendors.

A technology standard such as Q might be able to be modified, and a
device either supports the old way or the new way, but with SNMP it is
almost always a requirement that multi-vendor applications support both
ways, so the two ways cannot be in conflict (with very few
clearly-identified exceptions) or it can cause real interoperability
problems.

Having a separate mib for optional functionality helps to ensure that
mibs like the Q mib don't change, so people can count on it for
interoperability.

dbh


_______________________________________________
Bridge-mib mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/bridge-mib