Re: [AgentX] smilint messages for APM-MIB

[Bob Natale <Bob.Natale@AppliedSNMP.com>]

>At 8/6/2002:08:32 PM, Randy Presuhn wrote:

Hi Randy,

>> Date: Tue, 06 Aug 2002 17:11:49 -0700
>> From: Andy Bierman <abierman@cisco.com>
>> Subject: Re:  [RMONMIB] smilint messages for APM-MIB
>...
>> Should the RFC 1905 update increase the limit to the same value the
>> SMIng WG will use?  IMO, this CLR is too restrictive and has proven
>> to be too low a number considering current MIB design practices.
>> This CLR was added to SNMP to benefit MIB compiler and MIB engine
>> developers, at the expense of MIB designers. 
>...
>
>Although in my experience this rule actually complicated MIB
>compiler and SNMP engine implementation, and although I've
>never liked the limitation, I feel it's important to point
>out that RFC 2741 strongly assumes this limitation in the ad
>hoc encoding rules it employes.  For example, the AgentX pdu
>format only allocates eight bits to locate a sub-identifier.
>(I've BCC-ed the AgentX list so that folks there are aware of
>this discussion.)

Thanks.  An AgentX PDU encodes an OID value using an 8-bit
"m_subid" field (as part of a 4-byte OID-header element) to
record the number of sub-id elements in the OID, with the
corresponding number of 32-bit sub-id values following 
contiguously.  A comment in RFC2741 currently indicates
that the range of "m_subid" is "0-128".

I think AgentX could deal with this by modifying the elements
of procedure to say that if the "m_subid" value in the AgentX
OID header is less than 255 then it contains the number of sub-
id elements as before, but if the value is 255 then the first
32-bit value following the four OID header bytes specifies the
number of sub-id elements comprising the OID, the 32-bit values
for which then follow contiguously as before.

Of course, this would require a change to the specs that
would require re-certifying for interoperability (I suppose).

>Despite the number of WGs that this would impact, could we
>discuss it in just one place?  I really hate adding to the
>cross-posting list.

Agreed, I have bcc'd the AgentX list, just to make sure that
any interested parties there join in.


>If the proposal is to change the RFC 1905 update, I believe
>this belongs on the snmpv3 WG mailing list.

Agreed, but I've kept this reply to mibs@ops.ietf.org until
someone else makes the locus decision.

Cheers,

BobN