bandwidth estimation workshop
play

Bandwidth Estimation Workshop Mathieu Goutelle and Pascale Primet - PowerPoint PPT Presentation

tracerate : a non-intrusive method for measuring the hop-by-hop capacity of a path Bandwidth Estimation Workshop Mathieu Goutelle and Pascale Primet INRIA RESO team, LIP laboratory (ENS Lyon, France) 9-10 december 2003 Slides available at


  1. tracerate : a non-intrusive method for measuring the hop-by-hop capacity of a path Bandwidth Estimation Workshop Mathieu Goutelle and Pascale Primet INRIA RESO team, LIP laboratory (ENS Lyon, France) 9-10 december 2003 Slides available at http://perso.ens-lyon.fr/mathieu.goutelle/fichiers/sl_BEst2003.pdf Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 1/19

  2. Overview ■ Introduction ; Overview Introduction ■ Bandwidth measurement in IP networks ; BW mes. in IP networks The Packet Pair method Our proposition ■ The Packet Pair method ; Topology discovery Method principles ■ Our proposition: tracerate ; Data Analysis Capacity extraction ◆ Method principles ; Validations Accuracy study ◆ Data Analysis. Robustness study Experimental validation Utilization rate evaluation ■ Results and validations ; Conclusion References ■ Conclusion. Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 2/19

  3. Introduction ■ Simplicity of IP networks: no control channel, few informations Overview Introduction provided by equipments ; BW mes. in IP networks The Packet Pair method Our proposition Topology discovery Method principles Data Analysis Capacity extraction Validations Accuracy study Robustness study Experimental validation Utilization rate evaluation Conclusion References Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 3/19

  4. Introduction ■ Simplicity of IP networks: no control channel, few informations Overview Introduction provided by equipments ; BW mes. in IP networks The Packet Pair method Our proposition ■ Need of an external mean to evaluate the end-to-end Topology discovery Method principles performances: Data Analysis ◆ delay, loss rate: classical and easy ( ping , traceroute ); Capacity extraction Validations ◆ Capacity : Maximal available rate between two nodes ; Accuracy study Robustness study ◆ Available rate : Accessible rate between two machines given Experimental validation Utilization rate evaluation an utilization on the followed path ; Conclusion References Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 3/19

  5. Introduction ■ Simplicity of IP networks: no control channel, few informations Overview Introduction provided by equipments ; BW mes. in IP networks The Packet Pair method Our proposition ■ Need of an external mean to evaluate the end-to-end Topology discovery Method principles performances: Data Analysis ◆ delay, loss rate: classical and easy ( ping , traceroute ); Capacity extraction Validations ◆ Capacity : Maximal available rate between two nodes ; Accuracy study Robustness study ◆ Available rate : Accessible rate between two machines given Experimental validation Utilization rate evaluation an utilization on the followed path ; Conclusion References ■ Use of the delay between two machines → not enough to evaluate the duration of a data transfer. Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 3/19

  6. Bandwidth measurement in IP networks ■ A “rate” knowledge gives a more realistic view ; ◆ to estimate a transfer duration estimation ; ◆ to schedule transfer in grid computing ; ◆ to choose a data source or a data mirror ; ◆ ... Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 4/19

  7. Bandwidth measurement in IP networks ■ A “rate” knowledge gives a more realistic view ; ◆ to estimate a transfer duration estimation ; ◆ to schedule transfer in grid computing ; ◆ to choose a data source or a data mirror ; ◆ ... ■ Available rate : intrusive measurements ( iperf , MRTG , NWS ) or non-intrusive ( pathload [JD02]); Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 4/19

  8. Bandwidth measurement in IP networks ■ A “rate” knowledge gives a more realistic view ; ◆ to estimate a transfer duration estimation ; ◆ to schedule transfer in grid computing ; ◆ to choose a data source or a data mirror ; ◆ ... ■ Available rate : intrusive measurements ( iperf , MRTG , NWS ) or non-intrusive ( pathload [JD02]); ■ Total capacity : Method Type of measure Measure Protocol Receiver Variable Packet Size pathchar [Jac97] hop-by-hop slow UDP , ICMP no Packet Tailgating tailgater [LB00] end-to-end fast TCP , ICMP no Packet Pair pathrate [DRM01] end-to-end slow yes UDP Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 4/19

  9. Bandwidth measurement in IP networks ■ A “rate” knowledge gives a more realistic view ; ◆ to estimate a transfer duration estimation ; ◆ to schedule transfer in grid computing ; ◆ to choose a data source or a data mirror ; ◆ ... ■ Available rate : intrusive measurements ( iperf , MRTG , NWS ) or non-intrusive ( pathload [JD02]); ■ Total capacity : Method Type of measure Measure Protocol Receiver Variable Packet Size pathchar [Jac97] hop-by-hop slow UDP , ICMP no Packet Tailgating tailgater [LB00] end-to-end fast TCP , ICMP no Packet Pair pathrate [DRM01] end-to-end slow yes UDP ■ Issues : high-performance network, bottleneck localization, low intrusivity. Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 4/19

  10. The Packet Pair method (1) ■ Quite old principle (Van Jacobson, 1988 [Jac88]) ; Overview Introduction ■ A path is considered as a succession of delays (queue waiting BW mes. in IP networks The Packet Pair method time, transmission time, etc. ) ; Our proposition Topology discovery Method principles ■ Capacity evaluation through the dispersion (inter-packet delay) Data Analysis Capacity extraction measurement of two packets sent back-to-back ; Validations Accuracy study ■ This delay is the consequence of the smallest link on the path: Robustness study Experimental validation L/3c L/c Utilization rate evaluation Conclusion L/c References Sender Receiver C=3c C=c C=3c Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 5/19

  11. The Packet Pair method (1) ■ Quite old principle (Van Jacobson, 1988 [Jac88]) ; Overview Introduction ■ A path is considered as a succession of delays (queue waiting BW mes. in IP networks The Packet Pair method time, transmission time, etc. ) ; Our proposition Topology discovery Method principles ■ Capacity evaluation through the dispersion (inter-packet delay) Data Analysis Capacity extraction measurement of two packets sent back-to-back ; Validations Accuracy study ■ This delay is the consequence of the smallest link on the path: Robustness study Experimental validation L/3c L/c Utilization rate evaluation Conclusion L/c References Sender Receiver C=3c C=c C=3c ■ Hypothesis : No concurrent traffic! ■ Otherwise, concurrent traffic may cause the measure to under- or overestimate the real path capacity. Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 5/19

  12. The Packet Pair method (2) ■ Due to concurrent traffic, the measurement distribution is Overview Introduction multimodal [DRM01]: BW mes. in IP networks The Packet Pair method P={100,75,55,40,60,80}, L=Lc=1500B Our proposition P={100,75,55,40,60,80}, L=Lc=1500B Topology discovery 400 160 SCDR Method principles 360 Capacity Mode (CM) Data Analysis 140 # of measurements Capacity extraction # of measurements 320 Validations 120 280 Accuracy study u=80% 100 Robustness study 240 u=20% Sub−Capacity CM Experimental validation Dispersion Range 200 80 Utilization rate evaluation (SCDR) 160 Conclusion Post−Narrow 60 References 120 Capacity Mode 40 (PNCM) 80 PNCM 20 40 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Bandwidth (Mbps) Bandwidth (Mbps) ◆ under-estimation ( SCDR ): A packet has spaced the two probe packets ; ◆ over-estimation ( PNCM ): The first probe has waited for the second in an non-empty queue. Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 6/19

  13. Our proposition: tracerate ■ Objectives : to propose a method little intrusive to measure and to Overview Introduction localize the bottleneck of a path. It must work in a BW mes. in IP networks The Packet Pair method high-performance environment and without cooperation of the Our proposition Topology discovery destination ; Method principles Data Analysis Capacity extraction Validations Accuracy study Robustness study Experimental validation Utilization rate evaluation Conclusion References Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 7/19

  14. Our proposition: tracerate ■ Objectives : to propose a method little intrusive to measure and to Overview Introduction localize the bottleneck of a path. It must work in a BW mes. in IP networks The Packet Pair method high-performance environment and without cooperation of the Our proposition Topology discovery destination ; Method principles Data Analysis ■ Proposition : Capacity extraction Validations ◆ We use a Packet Pair because it is more robust regarding the Accuracy study Robustness study Experimental validation presence of invisible nodes [PDM03] ; Utilization rate evaluation ◆ We measure the hop-by-hop capacity (and delay and loss) up Conclusion References to the path bottleneck ; ◆ We eliminate “topology” parasitic modes with a better hop-by-hop knowledge of the topology (like traceroute ) ; ◆ We will be able to evaluate the hop-by-hop utilization rate up to the path bottleneck ; Mathieu Goutelle, 9-10 dec. 2003 Bandwidth Estimation Workshop 2003 – p. 7/19

Recommend


More recommend