Liaison statement
Reply to ITU-T SG12 LS - Harmonization of IP Capacity and Latency

State Posted
Submitted Date 2019-04-25
From Group ippm
From Contact Scott Mansfield
To Group ITU-T-SG-12
To Contacts A. C. Morton
ITU-T SG12 TSB
CcScott Mansfield
Bill Cerveny
IP Performance Measurement Discussion List
Mirja K├╝hlewind
Brian Trammell
Magnus Westerlund
itu-t-liaison@iab.org
Tommy Pauly
David Sinicrope
giulio.maggiore@TELECOMITALIA.IT
tsbsg11@itu.int
tsbsg12@itu.int
h.w.gierlich@head-acoustics.de
Response Contact Brian Trammell
Bill Cerveny
Tommy Pauly
Technical Contact scott.mansfield@ericsson.com
Purpose For information
Attachments (None)
Liaisons referred by this one LS - Harmonization of IP Capacity and Latency Parameters: Revision of Draft Rec. Y.1540 on IP packet transfer performance parameters and New Annex A with Lab Evaluation Plan
Body
IETF IPPM working group thanks ITU-T SG12 for your liaison describing

results from your evaluation of IP Capacity and Latency Metrics and

methods of measurement. The harmonization of these measurements

is welcome in our industry.


We have the following comments on the preliminary results, and the

evaluation plan:


After noting the efficacy of UDP measurements in your results, we

can state that UDP transport is the basis for most IPPM measurement

implementations, and IPPM's measurement protocols.


When assessing IP Capacity and Latency, there should be no attempt

to determine the technology involved in the path, such as AQM. The

method should measure the path as a black box.


The presence of packet-marking-sensitive technologies, such as

Diff-serv queues add complexity to IP Capacity and Latency.


There is not yet a metric that characterizes the BW-Delay product

in the IPPM-literature.


Live network testing may be affected by Network operator policy, it

could change from the lab measurements - the test traffic is part

of the background traffic. OTOH, the operator could be prioritizing

test traffic, especially their own authorized testing.


We also suggest to include QUIC-based measurements in your Lab

evaluation, if possible.


Where a model or post-processing needs to be applied to the

measurements, this aspect must also be specified, especially for transport

with multiple connections (averaging the connection performance).


Some of the IPPM WG participants who indicated their interest in

November 2019 have begun to share their comments informally, as

planned.


Please keep the IPPM WG informed of your progress, especially if there

are ways in which IPPM WG participants assist further.


Regards,

IPPM Chairs

Bill Cerveny (ietf@wjcerveny.com)

Brian Trammell (ietf@trammell.ch)

Tommy Pauly (tpauly@apple.com)