It was the original intention to generate a requirements document in WIMS
that would have covered the various specifications generated by the group.
It is generally more proper to agree on the requirements before the
specification is generated, although one gets better matching if the
requirements are generated after the specification is done. Indeed, with the
changing focus of WIMS and the changing constituency of the working group,
many of the original functional requirements have been dropped.
I also, perhaps, had the wrong idea of the purpose of the requirements
document, thinking that the requirements are to be primarily a statement of
why the standard is necessary, with some details of intended use. Taking the
Counter MIB requirements as a model, it appears that what is intended are
the requirements of the standard rather than the requirements for the
standard.
In either case, I agree that some sort of requirements document should be
issued and approved prior to submitting the Counter Spec for vote. I think
that a clear statement of why the standard is necessary and how it is to be
used would help significantly in getting PWG members to cast a vote as to
whether the proposed standard properly addresses that need.
Many of the last call comments came from working group members rather than
outside PWG members, and indeed some of the recent changes reflect a
constantly changing notion of what a particular figure or counter is
intended for. Perhaps better documented requirements would decrease the need
for this constant revision. Perhaps comments on the requirements would
provide either better understanding of the need for the number of counters,
or else would allow that number to be reduced.
I suggest that WIMS proceed with a requirements document covering the
abstract counters (which were generated as a simple alternative to a full
multi-function management structure), the manifestation of these abstract
counters in MIBS and XML schema, and the WIMS protocol itself.
Bill Wagner, Chairman WIMS WG
-----Original Message-----
From: owner-pwg at pwg.org [mailto:owner-pwg at pwg.org] On Behalf Of McDonald,
Ira
Sent: Monday, May 23, 2005 5:42 PM
To: 'thrasher at lexmark.com'; pwg at pwg.org
Subject: RE: PWG> SC topic - no requirements in Counter Spec or Port MIB
Hi,
I halfway agree with Jerry below.
I think it's possible to write pretty terse requirements
for the Port Mon MIB (because it has a small, focused
purpose in life).
But I tend to believe that the Counter spec needs more than
"a few paragraphs". For most specs (including PWG ones),
we have free-standing requirements documents that include
realistic use cases and detailed requirements derived from
those use cases. In the absence of WIMS (and certainly it
can't be referenced at all by the current Counter spec),
what conceivable use case is there for abstract counters?
Without any concrete mappings, the counters can't even be
transmitted.
There are three obvious use cases for counters:
(1) Monitoring basic network element health for operators;
(2) Monitoring usage and consumption for field service;
(3) Monitoring usage and consumption for billing/accounting.
Doing the first without strong mutual authentication is unwise,
but often done today internal to enterprise networks with
SNMPv1/v2 MIBs.
Doing the second without strong mutual authentication is likely
to lead to service contract disputes.
Doing the third without strong mutual authentication is simply
out of the question - unverifiable billing info is garbage.
That's not exhaustive, but _none_ of that is addressed in the
current Counter spec.
Cheers,
- Ira
Ira McDonald (Musician / Software Architect)
Blue Roof Music / High North Inc
PO Box 221 Grand Marais, MI 49839
phone: +1-906-494-2434
email: imcdonald at sharplabs.com
-----Original Message-----
From: owner-pwg at pwg.org [mailto:owner-pwg at pwg.org]On Behalf Of
thrasher at lexmark.com
Sent: Monday, May 23, 2005 3:55 PM
To: pwg at pwg.org
Subject: Re: PWG> SC topic - no requirements in Counter Spec or Port MIB
I would think that if it's really that hard to put down into words the
the problems being solved (and what's required to solve these
problems), then the resulting work product is probably not worth
of being standardized in the first place.
In the case of the PortMon MIB it should be pretty easy to
list what the problem is with respect to the configuration of a network
printer
port on the client side and the installation of an appropriate printer
driver, and
to create a bulleted list of what needs to be part of the MIB to enable
this.
In the case of the Counter Spec. it should also be fairly straightforward to
create a couple of paragraphs explaining what needs to be counted and
why as far as the need for billing/fleet management metrics.
I don't see this as a failure of the Process 2.0 since this was required in
the
original PWG Process document, I see it more as "our" failure to follow it.
JT
"McDonald, Ira" <imcdonald at sharplabs.com>
Sent by: owner-pwg at pwg.org
05/23/2005 03:19 PM
To: "'pwg at pwg.org'" <pwg at pwg.org>
cc:
Subject: PWG> SC topic - no requirements in Counter Spec or
Port MIB
Hi,
A topic for this Thursday's PWG Steering Committee:
Both the PWG Imaging System Counters spec and the PWG
Port Monitor MIB do _not_ have an internal requirements
section. Without an explicit variance from the PWG SC,
neither of these documents can enter Formal Approval
(the PWG Process/2.0 has no loopholes here).
On the other hand, the PWG Imaging Counter MIB _does_
have an internal requirements section.
Mea culpa - I should have realized this problem with
both specs a long time back, but the question is, what
to do?
Cheers,
- Ira
Ira McDonald (Musician / Software Architect)
Blue Roof Music / High North Inc
PO Box 221 Grand Marais, MI 49839
phone: +1-906-494-2434
email: imcdonald at sharplabs.com