[ippm] Re: [bmwg] Re: Joint BMWG/IPPM chartering

Hi,
发件人: Greg Mirsky [mailto:[email protected]]
发送时间: 2026年4月29日 6:56
收件人: [email protected]
抄送: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
主题: [bmwg] Re: [ippm] Joint BMWG/IPPM chartering

Hi Thomas,
Thank you for taking an interest in my notes and sharing your thoughts. Please find my follow-up notes below, tagged GIM2>>.

Regards,
Greg

On Mon, Apr 27, 2026 at 9:16 PM <[email protected]<mailto:[email protected]>> wrote:
Dear Greg,

Thank you very much for the feedback. As an individual, see my reply inline.

From: Greg Mirsky <[email protected]<mailto:[email protected]>>
Sent: Tuesday, April 28, 2026 1:16 AM
To: Graf Thomas, SCS-INI-NET-VNC-E2E <[email protected]<mailto:[email protected]>>
Cc: [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>
Subject: Re: [ippm] Joint BMWG/IPPM chartering


Be aware: This is an external email.


Hi Thomas,
thank you for sharing the updated proposals. I appreciate the work put into the proposal. I read the proposed charter for the new WG, please kindly consider my notes below:
·  The xxxx xxxx Working Group (xxxx WG) develops and maintains metrics, methodologies, and protocols that can be applied to the quality, performance, and reliability of data delivery services and applications running over public and private networks using IETF technologies.
While I agree that this is consistent with the IPPM WG's work, I cannot see how it applies to the scope of work performed in the BMWG. I have raised this concern several times: one of the major gaps between the two WGs is the environments for which the methodologies and procedures they specify are intended. The methodologies developed by the BMWG, AFAICS, from its current charter, are applicable in the lab environment, where an observer fully controls the network in all aspects, including configuration, offered payloads, and any changes to them. On the other hand, methodologies and protocols developed by the IPPM WG are intended for production networks, public or private, to provide operators with insights into network conditions with as little disturbance as possible to payloads. I don't think this statement closes the gap.

TG> I agree that this initial sentence doesn't address the difference between lab and production. I fully agree, this is, needs to be addressed, clearly in subsequent sentences. In my opinion, the "and" should be replaced with "or" in the " running over public and private networks" sentence.
GIM2>> I interpret "private networks" not as a synonym for "lab environment" as a private network may be a production network though not connected to the Internet. To me s/and/or/ in that sentence still leaves the domain studied by the BMWG outside of the scope of the new WG. Replacing "private" with "lab" might fix that.
[QW] I think the private network is still needed since it is the domain studied by the IPPM such as home network, office network, campus network, etc. I see lab environment/network as a special example of the private network with strict isolation with the outside of the network, controlled deterministic performance characteristics such as latency, packet loss to create baseline for benchmarking, and randomly configuration, etc.
GIM2>> After re-reading the list of the applicability, I notice that it misses an important metric that is measured only by methods developed by the BMWG - scalability. And this specificity so far has been outside of the scope for the work conducted by the IPPM WG.
[QW] I think it is related to resilience or availability, probably Benchmarking experts can come up with a different term.

TG> I think the following sentence should be change from

These metrics, methodologies, and protocols are designed as such that they can be used by network operators, end users, or independent testing groups in production or lab environments. Such metrics are intended to provide unbiased quantitative performance measurements.

to

These metrics, methodologies, and protocols are designed as such that they can be used by network operators and end users in production or independent testing groups in lab environments. Such metrics are intended to provide unbiased quantitative performance measurements.

Would that be better?
GIM2>>  It is, thank you.
·  Where in lab environments the testing procedures ensure consistent, reproducible, and reportable benchmarking results.
And what is produced in a real production network? Would someone will measure, for example, how fast an implementation learns 100000 prefixes in a live network? Or what's the rate of MAC flushes in EVPN that serve a subscriber? This is another manifestation of the gap I see between the objects of study for the IPPM and the BMWG. And I don't see that the gap was closed or even narrowed.

TG> You do have a point here. In my opinion, reflecting your comment, the charter proposal lacks a sentence that measurements in production can't be intrusive, interfere with production management, control and data plane. Therefore I suggest to add the following sentence right after "Where in lab environments" or before.

Where in production environments non-intrusive measurement procedures ensure that production management, control and data plane performance impact is as minimal as possible.

What do you think?
GIM2>> I think that truly non-intrusive, per RFC 7799, measurement methods are passive observations. I think that if "non-intrusive" is removed, the sentence you propose will still deliver the intended point.
[QW] Fully agree with this, since I also feel active measurement might introduce intrusive impact.

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