企业标识符下节点的SNMP MIB结构约定

Question

I have developed a Linux/Umbutu program running on what will be an imbedded P-based device. I wish for that piece of code to be able to send and receive the SNMP data for the entire product. So, I know the OID down to the company level, 1.3.6.1.4.1.34843, and I further know that the two SNMP enabled products we sell are 1.3.6.1.4.1.34843.1 and 1.3.6.1.4.1.34843.2. Pretty sure mine will be .3.

The question is, is there a known structure/convention for the nodes below this point? What I read seems to imply that I can use whatever structure I want. Would ...34843.1.0 exist? Is it a node to fetch (get) the list or max-count of same-level nodes? Would ...34843.1.0.0 exist? Would it make any sense? The guy who developed the first two MIBs is out of town for a while and I can't really ask him, and I'd like to make sure I understand his answer when he gives it to me.

Is there a document that describes the required/suggested organization? Probably some RFC, right. So, is there a document in a language spoken by humans?

Thanks in advance.

Answer 1

The most popular resource is the understanding SNMP MIBs book by Perkins and McGinnis. It describes MIBs in a more readable format. The Net-SNMP tutorials online might be helpful too, as they discuss a lot about MIBs. If you want the RFC, then RFC2578 is the right thing to read.

The quick hierarchy guidance would be something along the lines of:

1. Use 3 sections: one for objects, one for notifications, and one for conformance information
2. For notifications, root them at the .0 (see RFC3584 for details on why)
3. For each table, you might want a scalar showing the number of objects in a table and potentially a LastChanged object for showing how frequently configuration within the table might have changed).

There is a lot more guidance that can be found in the books and other sources listed above than can be repeated here, of course. RFCs 4181 , and 5249 may be helpful as well.