Minutes for XRBLOCK ======================================================================== Agenda IETF 82, Taipei 1520-1700 TUESDAY, November 15, 2011. Note taker: Al Morton, Varun Singh ------------------------------------------------------------------------ ** Action Items ******************************************************** - Chairs will update the milestones. Measurement-Identity - Author will include text about what kind of metrics should use the identity block and an explanation that drafts utilizing this draft (block) should include guidance that measurement blocks should be ignored when the identity block is missing. - The draft will go into WGLC after the revision. Delay metric - Won't wait until the one-way delay discussed in AVTCORE and move it ahead. - The draft will go into WGLC after the measurement-identity. Packet Delay Variation metric - Add 2 bits flag and write some text suggesting which type of reporting block should use this. - Ray will review the draft and provide comments. - The draft will go into the queue for WGLC after that. Burst/Gap related drafts - Make list of blocks/potential metrics and see what is the relation between the blocks and the usefulness (see if the metrics can be calculated easily out of pre-existing values etc.) - Discuss and close the issue relating to what to include and what to exclude. Run Length Encoding of Discarded Packets - Adopt the individual draft as a WG item. - Author submit the draft as a WG item. Multimedia Quality Metric Reporting - Discuss the issues about which MoS to include and whether IANA registry is needed. - Also discuss the way forward to merge the two drafts that are trying to address this milestone, will adopt the draft which WG thinks should be a baseline draft. ####################################################################### XRBLOCK Status Update presented by chairs ----------------------------------------------------------------------- - No comments. ####################################################################### Measurement Identity and information Reporting using SDES Presenter : Qin Wu Draft : draft-ietf-rxblock-rtcp-xr-meas-identity-01 ----------------------------------------------------------------------- ISSUE: Failure modes need to be documented if introducing a measurement information block. Colin Perkins - The draft describes how to correlate a metric with a period. Each XR draft should add one line that says that if they don't see the measurement identity block then they should ignore it. Colin.P - If the interval is different from the last reporting interval then they should use this draft. Roni.E/Colin.P - It should become a normative reference if the metric requires it. Roni.E - Isn't the previous block still applicable? maybe... Colin.P - Just covering the case that a measurement period is NOT received. Dan.R - Does this new RTCP SDES value become Mandatory to implement? *8 people in the room have read the draft. ??? - What about metrics that don't operate on measurement interval Colin.P - don't include it. Roni.E - Sequence numbers are not sufficient... Glen.Z - Is there a dependency for this draft on monarch and other blocks? Charles.E - tried to clarify the situation... Ready for WGLC? Roni.E - This draft needs to move forward first. ####################################################################### RTCP XR Report Block for Delay metric Reporting Presenter : Qin Wu Draft : draft-ietf-xrblock-rtcp-xr-delay-00 ----------------------------------------------------------------------- ISSUE: Do we need one way delay metric to this XRBLOCK? Roni.E - Can't make a one-way delay measurement if clocks are not synchronized. Need to know the accuracy required for this, then a suitable sync methodcan be used to achieve sufficiently accurate sync between the clocks. Colin.P - Repeated the above - need some other things before we can specify this... Need some help from TICTOC WG. Chairs - Do we want the dependency? Colin.P - Take it out and progress the rest. Glen.Z - This is just a reporting vessel - how to measure is somebody else's problem. Colin.P - Need a way to specify WHICH clock everybody is sync'd to. - Just need to sit down and write the text... Chairs - Tried to ask for interest, but few people have any Colin.P - Most of the draft is ready, but not one-way delay Chairs - Feel of the room: who cares about one-way delay? 1 person Colin.P - Take out one way delay and progress it, else hold it until one way delay progresses. Chairs - Will take it to WGLC once we decide what we are doing with one-way delay. ####################################################################### RTCP XR Report Block for Packet Delay Variation Metric Reporting Presenter : Qin Wu Draft : draft-ietf-xrblock-rtcp-xr-pdv-01 ----------------------------------------------------------------------- ISSUE: Do we need to be able to express Cumulative, Interval and Sample metrics? Do we want to apply the sample metrics only to PDV or to any future metric? Al.M - Sampled metrics must be adjacent for Inter-packet delay variation to be comparable to non-sampled results. MAPDV - not sure how sampling affest this ITU-T PDV is ok with sampling (reduced resolution of smaller sample). ITU-T PDV (and MAPDV, I believe) are one-sided (positive) so the negative threshold and neg. percentile are uneccessary (always zero) Al.M - Add 2 bits flag and write some text suggesting which type of block should use this. Colin.P - Let the drafts decide if they want to use it. If you want them to use it then define it. Alan.M - PPDV are absolute values, they are one-sided, reporting structure biased. Chairs - Ray volunteered to read the draft. Will be added to the WGLC after reflecting the changes and volunteer's comments. ####################################################################### RTCP XR Report Block for Discard metric, Burst/Gap Loss/Discard metric Reporting Presenter: Jing Zhao Drafts: draft-ietf-xrblock-rtcp-xr-discard-00, draft-ietf-xrblock-rtcp-xr-burst-gap-loss-00, draft-ietf-xrblock-rtcp-xr-burst-gap-discard-00, and ----------------------------------------------------------------------- ISSUE1: Combining loss/discard Move forward as proposed. (Do not combine the two but delete the overlap elements) ISSUE2: Early vs Late discard Charles.E: can you report both Early and late in the same interval? > Seemed like this issue was unresolved. Roni.E: Can run-length be used for calculating max burst Qin.W: Response to an open issue to keeping burst statistics (as opposed to the suggestions on the mailing list) Is there a need to add Max burst length into burst gap draft? Roni.E: Discuss with Varun and see if there is an overlap with his draft. Is there a need to add average burst? Roni.E: Need to make a list of all the potential metrics and see what can be calculated easily out of values one have. Roni.E: Make a list of blocks and see what is the relation between the blocks and the usefulness. Chairs: close open issues on the mailing list. ####################################################################### RTCP XR for Run Length Encoding of Discarded Packets Presenter : Varun Singh Draft : draft-ott-xrblock-rtcp-xt-discard-metrics-01 ----------------------------------------------------------------------- NO ISSUES presented. ? - What is the use-cases? Varun.S - For congestion control / Video etc. Varun.S - Not for monitoring. Chairs - Adopt it as a WG item. No opposition. ####################################################################### RTCP XR Blocks for multimedia quality metric Presenter : Glen Zohn Draft : draft-ott-xrblock-rtcp-xt-discard-metrics-01 ----------------------------------------------------------------------- ISSUE 1: De we need to distingish what MOS was used. YES ISSUE 2: Do we need to distinguish the channel where you collect the data? Hendri - The report has only use if it is in-band for passive monitors. Colin.P - If there are going to many MOS algorithms then make an IANA registry and do it out-of-band. If they are few then take a few bits in the RTCP report and list a table. What do you mean by channel? Glen/Qin: Need it for audio channel like left and right Colin: How do you distinguish between channels? or clearly specify a channel? Magnus: Score for the whole session or one channel? let the guys who want different channel MOS, let them specify it. Chairs: There are two drafts? who have read it? Only few raised their hands. Colin: they look similar except Alan's has more MoS metrics. Chairs: Please continue the discussion on the issues and discuss the way forward to merge the two drafts on the list. ------------------------ Time run out.