MIB-II on proxied chassis entities.

[[e-mail]]

Howdy,

I'd like opinions about the structure of MIB-II on proxied chassis entities.
For example, say the Chassis Entity Table contain the following entries:

	Agent:
	  chasEntityCommunity "public"
	  chasEntityIpAddress  16.21.16.128
	Proxied Entity 1:
	  chasEntityCommunity "lineCard1"
	  chasEntityIpAddress  16.21.16.128
	Proxied Entity 2:
	  chasEntityCommunity "lineCard2"
	  chasEntityIpAddress  16.21.16.128

so the agent is proxying for "lineCard1" and "lineCard2".

>From past discussions, the minimal MIB-II implementation has the system
and snmp groups.  One could:

  1) Keep a per proxied entity system group and a per proxied entity 
     snmp group.  So, the agent's snmp group counters count only 
     those messages directed at 16.21.16.128 on community "public".
     lineCard1's snmp group counters count only the snmp messages 
     directed at 16.21.16.128 on community "lineCard1".   Similarly, 
     lineCard2's snmp group counters count only those messages directed 
     and 16.21.16.128 on community "lineCard2".

     In other words, there is an individual snmp group per community.

  2) Keep a per proxied entity system group, and a per proxy agent 
     snmp group.  So, the agent's snmp group counters count all snmp
     messages directed at 16.21.16.128 regardless of community.
     The counters are the sum of all snmp traffic to "public", 
     "lineCard1", and "lineCard2".  The view of "public" contains
     the system, if, ip, etc., and snmp groups.  The view of
     "lineCard1" and "lineCard2" contains the system and snmp groups.

The later is easier to implement.  Is it less useful?  Is it more 
confusing to users?   Is one approach clearly better than the other?

						-Shawn