use of 0 as NULL in mtaGroupTable counters

develop!robb@sblab.att.com Wed, 18 August 1993 21:59 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa12649; 18 Aug 93 17:59 EDT
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa12645; 18 Aug 93 17:59 EDT
Received: from THOR.INNOSOFT.COM by CNRI.Reston.VA.US id aa25849; 18 Aug 93 17:59 EDT
Received: from INNOSOFT.COM by INNOSOFT.COM (PMDF V4.2-14 #1336) id <01H1W7CDP0OG984XLF@INNOSOFT.COM>; Wed, 18 Aug 1993 14:36:00 PST
Received: from att.att.com (att-out.att.com) by INNOSOFT.COM (PMDF V4.2-14 #1336) id <01H1W7C8USHS984Y8F@INNOSOFT.COM>; Wed, 18 Aug 1993 14:35:52 PST
Received: by develop (5.59/25-eef) id AA18671; Wed, 18 Aug 93 10:53:03 PDT
Date: Wed, 18 Aug 1993 10:53:03 -0700
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: develop!robb@sblab.att.com
Subject: use of 0 as NULL in mtaGroupTable counters
To: <>
Errors-to: ~ned+madman-errors@SIGURD.INNOSOFT.COM
Resent-message-id: <01H1W7CDR5UA984XLF@INNOSOFT.COM>
Message-id: <9308181753.AA18671@develop>
X-VMS-To: IN%"@sblab.UUCP:att!innosoft.com!ietf-madman"
MIME-version: 1.0
Content-type: text/PLAIN; CHARSET="US-ASCII"
Content-transfer-encoding: 7bit
Original-From: Marty Robb <develop!@develop.UUCP:robb>
Full-Name: Marty Robb
Content-length: 1483

The Mail Monitoring MIB states "the fact that all groups contain all possible
variables does not imply that all groups must use all possible variables...
In this sort of configuration all unused counters would be left at zero."

I can't really take credit for the following comment.  Stallings in
"SNMP, SNMPv2 and CMIP: The Practical Guide to Network-Management Standards"
(p. 166) states:
   "it may be important for the management station to know whether the count is
    actually 0 or whether the agent simply isn't doing the counting"

Since Stallings got flamed for his (I think excellent) book, let me follow up
with his quote from Steve Waldbusser:
   "I would be very angry if the following happened to me:  While debugging
    a packet loss problem, I inspect ifInErrors on an interface and find that
    the interface has not received any error packets...later, I learn from my
    system vendor that ifInErrors is not implemented, but is stating anyway
    that no errors have been received."

Its easy to imagine implementors of the Mail Monitoring MIB that find it
convenient to calculate the *Submitted* and *Delivered* variables while
leaving the *Stored* variables zero.  That's very close to Steve Waldbusser's
example.  If it is unacceptable to return noSuchName errors for NULL values,
can we at least choose a better NULL value than 0?  For example, HIGH VALUES
for Counter would be 4,294,967,295.

Regards,
Martin Robb
AT&T Santa Barbara Lab
robb@sblab.att.com