[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: some more comments on draft-ietf-psamp-sample-tech-05.txt

Hi Benoit,

see comments inline

Benoit Claise wrote:
Dear all,

In order to produce the next version of the PSAMP MIB, I've reviewed the "sampling and filtering techniques for IP packet selection" draft... this time with a PSAMP MIB hat. I still have a few comments on the draft.

1. section 2 "PSAMP Documents Overview"
 [PSAMP-MIB]  "Definitions of Managed Objects for Packet Sampling" describes the PSAMP management Information Base.
I see that the current MIB description use the same definition. However, the MIB specifications go well beyond sampling... there is also filtering and hashing!
What about "definitions of managed objects for sampling and filtering techniques for IP packet selection"?
   - the draft "A Framework for Packet Selection and Reporting" should be changed accordingly
   - the MIB definition should be changed accordingly

According to out definition sampling and filtering are both packet selection techniques So in my opiniton the MIB title should change to "Definitions of Managed Objects for Packet Selection",  the FW document title can remain.

2. "Packet Content" definition

done (changed for new version)
3. If I recall correctly, the following terms have been changed in the latest "A Framework for Packet Selection and Reporting" (at least I recall some discussions on the mailing list): population size, sample size, configured selection fraction, attained selection fraction
Nick and I had a teleconferecne some time ago and agreed on those terms. So they are consistent in both drafts.

4. Table in section 4
Change "Filter" to "Filtering" in the category column
Same section, change "cascade of a filter and a Sampling scheme" to "cascade of Filtering and Sampling schemes"

5. Section
Still non-ascii characters with "sample and hold" and "local export"
6. Section
"In the former case determines pseudorandom variates rather than selection probabilities" ??

7. Section 6.3 "Router State Filtering"
Remove the OR.

done (now only AND is possible)
Since it seems no problem for the MIB or implementation to also have a NOT we could also allow a NOT for the matching filter combination
Thomas, Benoit do you agree ?

8. Section 7
"In order to be compliant with PSAMP, it is sufficient to implement one of the proposed schemes".
As far as I recall, this will be a standard track document, so it should be written in a more formal way.
"In order to be compliant with PSAMP, one least of the proposed scheme MUST be implemented"
9. section 7.1
SELECTOR_ID: ... the ID can be calculated under consideration of the ASSOCIATIONS and a local ID.
What does the "under consideration" mean?
To get a unique ID both need to be conisdered (local ID and associations ) ==> changed the text to "calculated as combination of..."

10. section 7.1
case non-uniform probabilistic and case flow state both refer to section
However, both the method descriptions and the section don't clearly explain which SELECTOR_PARAMETERS we need
Yes, because we decided that those techniques are not mature enough nowadays to define generic parameters for the scheme and this is stated in

11. section 7.1
"The ASSOCIATIONS field describes the Observation Point and (possibly) the IPFIX processes"
Possibly -> optionally without parentheses.
Note: the MIB should take care that these parameters are optional. Proposal: a value of 0 means unspecified?
changed. 0 (unspecified) than means it should apply to all IPFIX processes at this observation point ?

12. section 7.1
Case Matching -> this is not too clear that these are pairs of (field, value). And that in case of multiple match criteria, we have several "case matching" bound by a logical AND.
BTW, IPFIX speaks of information element and not field, so I guess it should be a (information element, value) pair.
changed and sentence added
13. section 7.1
case hashing -> this is not too clear how the parameters in there match the definitions of hash domain, hash range, hash selection range, hash-based selection, etc... BTW I have the same problem with the MIB definitions (which obviously match these definitions)

14. section 7.1
"all router state entries can be linked by AND, OR, NOT operators" ->to be limited by AND

done. Maybe possible to allow NOT operators ?
15. there are still a lot of capitalized terms before definition. The list is too long is to described in an email.
 Either use the find function, either call me as I marked them down on the paper draft I reviewed.


Regards, Benoit.

to unsubscribe send a message to psamp-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/psamp/>

Dipl.-Ing. Tanja Zseby			    	      	
Fraunhofer Institute FOKUS			Email: zseby@fokus.fraunhofer.de	
Kaiserin-Augusta-Allee 31			Phone: +49-30-3463-7153
D-10589 Berlin, Germany				Fax:   +49-30-3463-8153
"Living on earth is expensive but it includes a free trip around the sun." (Anonymous)