Dear Qin,

As one of the authors. Thanks a lot for the review.

Regarding point 4 and 6. This has already being raised by Paul at 
https://mailarchive.ietf.org/arch/msg/opsawg/zAyBkk5I_3SKHvePBjugmFvXlvg/ and 
is in the queue for -18 revision

https://author-tools.ietf.org/diff?doc_1=draft-ietf-opsawg-ipfix-on-path-telemetry-17&url_2=https://raw.githubusercontent.com/network-analytics/draft-ietf-opsawg-ipfix-on-path-telemetry/refs/heads/main/draft-ietf-opsawg-ipfix-on-path-telemetry-18.txt

Please let me know wherever it addresses also your points or not.

Regarding point 5. The reason for choosing unsigned64 for 
pathDelaySumDeltaMicroseconds is because unsigned32 would limit the amount of 
accounted flows to approx 100'000 if in average the delay for a flow would be 
42 milliseconds. Since the standard flow timeout setting for active flows is 
usually set to 60 seconds, that would lead to approx. 1666 flows per second. 
Therefore the authors believed unsigned64 might be the better choice. For the 
other IPFIX entities since it represents one measurement, not the sum, it means 
that a measurement for up to 4294 seconds can be exported. In reality 
measurement above one second is very unrealistic. Especially considering that 
this is most likely only applied in a constraint network environment.

Regarding point 7. The same data set can be applied for both, 
pathDelayMeanDeltaMicroseconds and pathDelaySumDeltaMicroseconds. As described 
in 
https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-on-path-telemetry-17#section-7.2,
 the idea behind pathDelaySumDeltaMicroseconds is to offload the IPFIX exporter 
from the mean calculation. Therefore from an implementation perspective, 
realistically an implementer is choosing either pathDelayMeanDeltaMicroseconds 
or pathDelaySumDeltaMicroseconds. Technically pathDelayMeanDeltaMicroseconds 
and pathDelaySumDeltaMicroseconds can be exported both. Should this be 
emphasized more in the Operational Considerations section?

On point 1-3, we will get back in a separate email.

Best wishes
Thomas

-----Original Message-----
From: Qin Wu via Datatracker <[email protected]> 
Sent: Saturday, June 7, 2025 12:24 PM
To: [email protected]
Cc: [email protected]; [email protected]; 
[email protected]
Subject: draft-ietf-opsawg-ipfix-on-path-telemetry-17 ietf last call 
Perfmetrdir review


Be aware: This is an external email.



Document: draft-ietf-opsawg-ipfix-on-path-telemetry
Title: Export of Delay Performance Metrics in IP Flow Information eXport (IPFIX)
Reviewer: Qin Wu
Review result: Ready

This document specifies 4 new IPFIX IEs for on path delay measurement results 
reporting. I have read the latest version of this draft and I think the draft 
is on the right track. Here are a few comments I would like to ask authors to
consider:

1. Section 3.2.5 T0 and T1 definitions
"
   T0:  T time, the start of a measurement interval (format "date/time"
      as specified in Section 5.6 of [RFC3339]; see also "date-and-time"
      in Section 3 of [RFC6991]).  The UTC Time Zone is required by
      Section 6.1 of [RFC2330].  When T0 is "all-zeros", a start time is
      unspecified and Tf is to be interpreted as the duration of the
      measurement interval.  The start time is controlled through other
      means.

   Tf:  A time, the end of a measurement interval (format "date/time" as
      specified in Section 5.6 of [RFC3339]; see also "date-and-time" in
      Section 3 of [RFC6991]).  The UTC Time Zone is required by
      Section 6.1 of [RFC2330].  When T0 is "all-zeros", an ending time
      and date is ignored and Tf is interpreted as the duration of the
      measurement interval.
"
Both T0 definition and T1 definition emphasize when T0 is "all-zeros", Tf is 
interpreted as the duation of the measurement interval, I think it is not 
necessary to repeat this statement, suggest to remove such duplication in both 
defintions.

2. Measurement method in section 3.4.2.1, Section 3.4.2.2, section 3.4.2.3, 
section 3.4.2.4 Since section 3.4.2.3 provide calculation formula for 
OWDelay_HybridType1_Passive_IP_RFC[RFC-to-be]_Seconds_Max, for consistency, I 
think section 3.4.2.1, section 3.4.3.2, section 3.4.2.4 should also provide 
calculation formula for mean, min and sum related metric.

3. Measurement method for section 3.4.2.4 Section 3.4.2.4 said:
"
However in this case FiniteDelay or MaxDelay MAY be used.
"
Can you clarify in which circumstance the finitedelay or maxdelay will be used?
Also I am wondering whether you use MinDelay or MeanDelay to calculating 
statistics, see section 4.2.5 and section 4.3.5 of RFC6049 for calculating 
formula.

4. Table 2
"
+-----------+-----------+------+-------------+-------------+-----------+
| ingress   | egress    | Node | destination | srhActive   |   Path    |
| Interface | Interface |      | IPv6Address | SegmentIPv6 |   Delay   |
+-----------+-----------+------+-------------+-------------+-----------+
|    271    |    276    |  R1  | 2001:db8::2 | 2001:db8::4 |    0 us   |
+-----------+-----------+------+-------------+-------------+-----------+
|    301    |    312    |  R2  | 2001:db8::3 | 2001:db8::4 |   22 us   |
+-----------+-----------+------+-------------+-------------+-----------+
|    22     |     27    |  R3  | 2001:db8::4 | 2001:db8::4 |   42 us   |
+-----------+-----------+------+-------------+-------------+-----------+
|    852    |    854    |  R4  | 2001:db8::4 | 2001:db8::4 |  122 us   |
+-----------+-----------+------+-------------+-------------+-----------+
           Table 2: Example table of measured delay. Ascending by delay.
"
Consider the example depicted in figure 1, I am wondering whether the 
destinationIPv6 addresses for both Node R3 and Node R4 should be the same, 
i.e., 2001::db8::4, can you clarify this?

5. Section 7.3
Can you clarify why Unsigned64 has been chosen as type for 
pathDelaySumDeltaMicroseconds while Unsigned32 has been choosen as type for 
IPFIX Information Elements? why not choose unsigned64 or unsigned32 for all new 
defined IPFIX IEs?

6. Section 3.1.2
The abbreviation of Observation point (OP) has been repeated multiple times, 
also observation point is defined in RFC7011, I don't see RFC7011 defines 
abbreviation for Observation point, therefore I suggest to remove such 
abbreviation as well. Sometime use OP introducing confusing, e.g., quoted the 
following text in section 3.2.1 "  With the OP [RFC7011] typically located 
between the hosts
   participating in the IP Flow, the one-way delay metric requires one
   individual measurement between the OP and sourcing host "
For the first reader, he will not understand what "OP" is and also there is no 
OP abbreviation defined in RFC7011.

7. Aggregated On-Path Delay Examples in Appendix Can I use the same Data Set 
encoding to carry both pathDelayMeanDeltaMicroseconds and 
pathDelaySumDeltaMicroseconds? Or you think I should use two different Data Set 
encoding for pathDelayMeanDeltaMicroseconds and pathDelaySumDeltaMicroseconds 
respectively?



Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

_______________________________________________
OPSAWG mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to