Delay Variation Applicability Statement draft-morton-ippm-delay-var-as-02 March 21, 2007 Al Morton Benoit Claise “

Page 2 Recap of activity since IETF-67 Benoit Claise read the draft (through the early comparison section) and shared comments with Al off-list. Al summarized those comments on list, and produced the -01.txt version (Nov 27, 2006). Benoit completed his review and summarized all his comments on the list (Dec 1, 2006). Al and Benoit discussed these comments over the next few months. -02.txt version March 4, 2007 with Benoit as co- author.  much new material and questions for IPPM WG discussion

Page 3 Applicability Problem: “How will the results be used?” Krzanowski introduced the Delay Variation Problem at IETF-64 “How” Question asked at IETF-65, no suggestions yet RFC 3393 lists two key uses for the Delay Variation Metric Memo Considers these 2 Tasks and 2 Special Circumstances SRC DST Network Characterization: Inferring Queue Occupation on a Path Application Performance Estimation: Sizing of De-Jitter Buffers What about Spatial Composition ? What if circumstance s are challenging?

Page 4 Modified Outline of the Draft 1. Introduction 1.1. Background Literature in IPPM and Elsewhere 1.2. Organization of the Memo Purpose and Scope Brief Descriptions of Delay 3. Variation Uses (four) 4. Formulations of IPDV & PDV Earlier Comparisons Additional Properties and Comparisons 6.1. Packet Loss 6.2. Path Changes Lossless Path Change Path Change with Loss 6.3. Clock Stability and Error 6.4. Spatial Composition 6.5. Reporting a Single Number 6.6. Jitter in RTCP Reports 6.7. MAPDV Load Balancing Applicability of the Delay Variation Forms and Rec 7.1. Uses 7.2. Challenging Circumstances Clock Issues Frequent Path Changes Frequent Loss Load Balancing Measurement Considerations for Vendors, Testers, and Users 8.1. Measurement Stream Characteristics 8.2. Measurement Units 8.3. Test Duration 8.4. Clock Sync Options 8.5. Distinguishing Long Delay from Loss 8.6. Accounting for Packet Reordering 8.7. Results Representation and Reporting … Appendix on Reducing Delay Variation in Networks Appendix on Calculating the D(min) in PDV 13.

Page 5 Inter-Packet Delay Var. (selection f = previous packet) Src Dst T1 R1 T2 R2 T3 R3 T4 R4 T5 R5 IPDV(2) = (R2-R1) – (T2-T1) (5)(2)(4) IPDV(4) = (R4-R3) – (T4-T3) (3)

Page 6 Packet Delay Variation (selection f = minimum delay pkt in stream) Src Dst T1 R1 T2 R2 (minimum) T3 R3 T4 R4 T5 R5 PDV(3) = (R3-T3) – (R2-T2) (2)(4) PDV(4) = (R4-T4) – (R2-T2) (1,3,5)

Page 7 Summary of Comparisons Challenging Circumstances for measurement:  IPDV form offers advantages when  Path changes are Frequent  Meas. System Clocks exhibit significant Skew  PDV form is less sensitive to Packet Loss Spatial Composition of DV metric:  All known methods use PDV,  IPDV sensitivity to sequence is an issue Estimate of Queuing Time & Variation:  PDV estimates this, especially when sample min = true min Determine De-jitter Buffer Size Required  PDV “pseudo-range” reveals this property by anchoring the distribution at the minimum delay

Page 8 Examples from IPPM List Discussion (2000) Ex. A – Smooth Trend IPDV Range = (-10, 10) PDV Range = ( 0, 50 ) Ex. B – Faster Var with loss IPDV Range = (-20, 40) PDV Range = ( 0, 50 ) How can the detailed information IPDV provides be used? (to complete some task)

Page 9 What if there is load-balancing in the network? Load- balancing is based on the IGP metrics, while the delay depends on the path. So, we could have a multi-modal distribution, if we send test packets with different characteristics such as IP addresses/ports. Should the delay singletons using multiple addresses/ports be combined in the same sample? The PDV form makes the identification of multiple modes possible. Should we characterize each mode separately?  This question also applies to the Path Change case.

Page 10 New Sections Section 8: Measurement Considerations for Vendors, Testers, and Users  This section provides guidance in many areas – Useful?  Do we want to recommend measuring BOTH IPDV and PDV?  Do we want to reconstruct the IPDV and DV later on, with a different interval? Appendix: Guidance for reducing DV in networks  Useful? Appendix: Calculation of the Minimum Delay, D(min)  how is this D(min) calculated? Is it DV(99%) as mentioned by Roman in  do we need to keep all the values from the interval, then take the minimum?  Or do we keep the minimum from previous intervals?

Page 11 Summary IPDV and PDV each have their Strengths and Weaknesses Suggestions for additional Tasks & Circumstances Need Feedback on newly proposed sections and questions raised in the text of the draft.