Re: [IPFIX] IPFIX processId uniqueness
Brian Trammell <[email protected]>
| Newsgroups | gmane.ietf.ipfix |
|---|---|
| Message-ID | <[email protected]> |
Hi, Paul, These appear to me to be completely implementation-dependent, and are used to scope metadata reporting within an observation domain. I can't think of a useful application of comparison among epId and mpId, so I'm not sure why you'd need to enforce process ID uniqueness across multiple process ID IEs. Because of the implementation-specificness, the CP, for example, shouldn't presume that an MP an EP are the same (host OS) process just because their epId and mpId match, but that's okay because I can't see why the CP would ever have to deduce that anyway. Regards, Brian On 13 Mar 2013, at 7:26, Paul Aitken <[email protected]> wrote: > Dear IPFIX experts, > > RFC5102 defines two processId IEs: meteringProcessId and exportingProcessId. And we could expect more *ProcessId IEs in future. > > Are these expected or required to be unique? > > Do we allow that a single process can both meter and export, therefore they are not unique? > > What about the case of multiple CPUs, where although the processes are unique, the processIDs happen to be identical? > > Thanks, > P. > _______________________________________________ > IPFIX mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/ipfix _______________________________________________ IPFIX mailing list [email protected] https://www.ietf.org/mailman/listinfo/ipfix