Re: query on draft-ietf-iplpdn-frmib-dte-04.txt

[John Shriver <[e-mail]>]

I suppose the primary restriction of modeling a FR interface as one
SNMP interface is that, for instance, in MIB-II you can't correlate
which IP address is assigned to a particular DLCI (or DLCI's), since
they include one ipAdEndIfIndex.  (I presume the same applies to
Novell's IPX MIB, someone borrowed my NLSP standard yesterday.)

The similar restriction applies to ipRouteIfIndex.  There is a loss of
information. 


Now, maybe our conceptual problem (at Shiva) is that we didn't read
the description below correctly:

    frCircuitIfIndex OBJECT-TYPE
        SYNTAX   Index
        MAX-ACCESS   read-only
        STATUS   current
        DESCRIPTION
           "The ifIndex Value of the ifEntry this  virtual
           circuit is layered onto."
       ::= { frCircuitEntry 1 }

Does this mean that frCircuitIfIndex must be one of the values of
frDlcmiIfIndex?  Or can it be a whole row in the SNMP interface table,
distinct from any value of frDlcmiIfIndex, and you figure out what
physical interface it is by the RFC 1573 ifStackTable.  Perhaps that
is where we went astray reading the RFC & I-D here.

If this is the case, two examples and some text in the RFC might make
things much more obvious:

Case 1: One ifEntry per FR interface:

frDlcmiIfIndex	...
1
2

frCircuitIndex	frCircuitDlci	...
1		88
1		201
1		222
2		72
2		99

(No need for ifStackTable.)


Case 2: One ifEntry per FR circuit (or circuits):

frDlcmiIfIndex	...
1
2

frCircuitIndex	frCircuitDlci	...

10		88
11		201
12		222
12		223
12		224
20		72
21		99

ifStackHigherLayer	ifStackLowerLayer
10			1
11			1
12			1
20			2
21			2