RE: question on key localizaton
Bob Natale <[email protected]>
| Newsgroups | gmane.ietf.snmpv3 |
|---|---|
| Message-ID | <[email protected]> |
At 11/19/2002:12:21 AM, ??? wrote: Hi "???", >dp> Once you correctly fill out the above, the answers to your questions >dp> will become visible. > >However, I went through all these with my original assumption, > >"A non-authoritative engine Ea know at least one of the security Name and >its corresponding auth key and priv key on the authoritative engine Eb if >and only if Ea can send get, getnext, getbulk, set and inform requests (so >called, Confirmed Class) securely to Eb." Well, while it's probably just a nit, you must mean "...or inform requests" (not "and"). This distinction is important both for the precise statement of your assumption and also for the (related) recognition that in SNMPv3 we no longer distinguish applications between "agents" and "managers", allowing any sensible mix of the five SNMP functionality types (mis-identified, IMHO, as "applications" by RFC2573) in any single application. However, I believe that your reference to "Confirmed Class" there indicates the rationale sufficiently. Forgetting discard and engine-level Report scenarios for the moment, a secure confirmed class PDU (i.e., everything except Response, Trap, and Report) must carry appropriate security credentials, which must be known a priori. Means of configuring that security information in the PDU originator must be provided by the overall application environment. I think that is a correct description of the relevant logic of SNMPv3 security design. Whether the logic itself is "correct" (it's certainly workable) may be a matter of personal choice. :-) >which is also what I wanted to verify and nothing is so obvious to me about >this assumption. How can an assumption become obvious with questions which >can be answered only when we admit it? <weak humor warning!> I'm not sure if you are new to SNMP or not -- your questions and observations are very valid -- but it's clear that you are new to the SNMP e-mail list! We have many interesting ways of beating a resolution out of a problem. </weak humor warning!> Cheers, BobN <aka, "An equally guilty party">