Measuring YouTube from Dual-Stacked Hosts Appendix from - - PowerPoint PPT Presentation

▶
measuring youtube from dual stacked hosts
SMART_READER_LITE
LIVE PREVIEW

Measuring YouTube from Dual-Stacked Hosts Appendix from - - PowerPoint PPT Presentation

Oct 23, 2022 37 likes •289 views

Measuring YouTube Google Global Caches March 2015 Jacobs University Bremen, Bremen, Germany Computer Networks and Distributed Systems Group, Aalto University, Espoo, Finland School of Electrical Engineering, Jrgen Schnwlder Jrg Ott

slide-1
SLIDE 1

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Measuring YouTube from Dual-Stacked Hosts

Passive and Active Measurement (PAM) Conference, New York

Saba Ahsan

saba.ahsan@aalto.fi

Vaibhav Bajpai

v.bajpai@jacobs-university.de

Jörg Ott

jorg.ott@aalto.fi

Jürgen Schönwälder

j.schoenwaelder@jacobs-university.de School of Electrical Engineering, Aalto University, Espoo, Finland Computer Networks and Distributed Systems Group, Jacobs University Bremen, Bremen, Germany

March 2015

Leone FP7 EU Project: leone-project.eu 1 / 25

slide-2
SLIDE 2

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Introduction | Motivation

▶ Large IPv6 broadband rollouts1 since World IPv6 Launch Day in 2012. ▶ Increased global adoption of IPv6 to 6% (as seen by Google as of March 2015). ▶ Studies show how YouTube contributes heavily to volumes of IPv6 traffjc [1]:

−6 −5 −4 −3 −2 −1 W6D + 1 + 2 + 9 +10 +11 +12 +16 +17 +18 +19 +20 +21 +22 unknown well−known ftp−data ssh ssl dns http Percent of volume 20 40 60 80 100

  • Fig. 7. Application mix per day for all IPv6 traffic (campus).

−6 −5 −4 −3 −2 −1 W6D + 1 + 2 + 9 +10 +11 +12 +16 +17 +18 +19 +20 +21 +22

  • ther

edu gov Facebook

  • pen source

Google YouTube Percent of volume 20 40 60 80 100

  • Fig. 8. Daily HTTP mix (campus).

1Comcast, Deutsche Telekom AG, AT&T, Verizon Wireless, T-Mobile USA 2 / 25

slide-3
SLIDE 3

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Introduction | Research Question Do users experience benefjt (or an added penalty) when streaming YouTube videos over IPv6?

3 / 25

slide-4
SLIDE 4

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Introduction | Research Contributions

  • 1. TCP connect times to YouTube makes Happy Eyeballs [2] prefer IPv6.
  • 2. Lower throughput is achieved when streaming YouTube over IPv6.
  • 3. YouTube content caches over IPv6 are largely absent.

To the best of our knowledge, this is the fjrst study to compare YouTube performance

  • ver IPv4 and IPv6 from dual-stacked networks.

4 / 25

slide-5
SLIDE 5

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology

5 / 25

slide-6
SLIDE 6

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | YouTube Performance Test

  • 1. Takes a YouTube URL as input and scrapes the downloaded HTML page.
  • 2. Extracts container formats2, available resolutions and media server locations.
  • 3. Locally resolves DNS names of media server locations.
  • 4. Establishes concurrent TCP connections for audio and video streams.

▶ Measures TCP connect times by recording connect(…) call completion time. ▶ DNS resolution time is not accounted.

  • 5. Fetches audio and video streams over concurrent HTTP sessions.

▶ Ensures temporal synchronization between audio and video streams. ▶ Measures throughput achieved over the single TCP connection for each stream.

  • 6. Extracts frame timestamps from container to mimic a playout.

▶ A 2 second prebufgering is applied before starting playout timer. ▶ Measures stall duration whenever a frame fails to arrive before its playout time. ▶ A stall triggers 1 second of rebufgering before resuming playout timer. 2Tie YouTube test supports three container formats: MP4, WebM and FLV 6 / 25

slide-7
SLIDE 7

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | Speed Test

▶ Measures achievable throughput over the line. ▶ Uses 3 simultaneous TCP connections to fetch 1 GB, non-zero, binary fjle. ▶ HTTP GET request is made to the nearest (based on latency) M-Lab server. ▶ Detailed in the SamKnows test suite [3] description. ▶ We modifjed the test to also enable measurements over IPv6. ▶ We use these line rates to baseline our YouTube throughput measurements.

7 / 25

slide-8
SLIDE 8

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | Selection of YouTube Videos

▶ We use the YouTube v3 API3 to prepare a list of globally popular videos where:

  • 1. Video duration > 60s.
  • 2. Video is available in Full HD format.
  • 3. Video has no regional restrictions.

▶ List is refreshed every 12h on the SamKnows backend. ▶ Each probe pulls this list on a daily basis.

3https://developers.google.com/youtube/v3/docs/videos/list 8 / 25

slide-9
SLIDE 9

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | Selection of Video Bitrate

▶ YouTube provides a list of available resolutions (and their required bitrates). ▶ Tie YouTube test currently does not support DASH [4]. ▶ We use speed test results to limit maximum bitrate. ▶ We also support 2 operation modes:

  • 1. Non-adaptive mode.

▶ Test downloads the same video resolution despite stalls. ▶ Although, does not mimic the behavior of YouTube players. ▶ However, still useful to compare IPv4 vs IPv6 performance in identical conditions.

  • 2. Step-down mode.

▶ Test steps down to a lower video resolution on a stall. ▶ Portrays a more user-oriented behaviour. 9 / 25

slide-10
SLIDE 10

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | Measurement Setup

DSL/Cable Modem SamKnows Tests results YouTube Servers YouTube Media Servers YouTube client 2.Media URLs 3.Audio Stream 3.Video Stream SamKnows Backend

Probe

4.HTTPS POST 1.HTTP(S) GET

▶ YouTube test runs every hour (once for IPv4 and subsequently for IPv6). ▶ Speed test runs every 6 hours (once for IPv4 and subsequently for IPv6).

10 / 25

slide-11
SLIDE 11

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Methodology | Measurement Trials

# LOCATION PROVIDER TYPE #01 BREMEN KABELDEUTSCHLAND RESIDENTIAL #02 BREMEN DEUTSCHE TELEKOM RESIDENTIAL #03 STOCKHOLM SITAB RESIDENTIAL #04 FUKUOKA ASAHI NET RESIDENTIAL #05 MADRID JAZZ TELECOM RESIDENTIAL #06 ALLEUR EDPNET RESIDENTIAL #07 BREMEN DEUTSCHE TELEKOM RESIDENTIAL #08 SHIZUOKA BIGLOBE NEC RESIDENTIAL #09 CERN CERN RESEARCH #10 BREMEN DFN NREN #11 TIMISOARA ROEDUNET NREN #12 LOUVAIN BELNET NREN #13 BREMEN DFN NREN #14 HELSINKI FUNET NREN #15 LONDON BSKYB-BROADBAND LAB #16 TORINO TELECOM ITALIA LAB #17 MADRID BT ESPANA LAB #18 IPSWICH BT UK LAB #19 NIIGATA NDAC IXP #20 BRAUNSCHWEIG GAERTNER DATENSYSTEME BUSINESS #21 OLTEN INIT SEVEN BUSINESS 11 / 25

slide-12
SLIDE 12

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis4

4Tie results are derived from measurements conducted for 20 days in September 2014. 12 / 25

slide-13
SLIDE 13

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | Speed Tests (Residential)

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA01

0.00 0.25 0.50 0.75 1.00 3 6 9 12

Throughput (Mbps) CDF

MA02

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA03

0.00 0.25 0.50 0.75 1.00 5 10 15

Throughput (Mbps) CDF

MA04

version IPv4 IPv6

0.00 0.25 0.50 0.75 1.00 0.0 0.5 1.0 1.5

Throughput (Mbps) CDF

MA05

0.00 0.25 0.50 0.75 1.00 10 20 30

Throughput (Mbps) CDF

MA06

0.00 0.25 0.50 0.75 1.00 10 20 30 40 50

Throughput (Mbps) CDF

MA07

0.00 0.25 0.50 0.75 1.00 5 10 15

Throughput (Mbps) CDF

MA08

13 / 25

slide-14
SLIDE 14

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | Speed Tests (Research, Business)

0.00 0.25 0.50 0.75 1.00 25 50 75

Throughput (Mbps) CDF

MA09

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA10

0.00 0.25 0.50 0.75 1.00 25 50 75

Throughput (Mbps) CDF

MA11

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA12

version IPv4 IPv6

0.00 0.25 0.50 0.75 1.00 100 200 300 400

Throughput (Mbps) CDF

MA13

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA14

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA20

0.00 0.25 0.50 0.75 1.00 25 50 75 100

Throughput (Mbps) CDF

MA21

14 / 25

slide-15
SLIDE 15

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | TCP Connect Times

# 1 # 2 # 3 # 4 # 5 # 6 # 7 # 8 # 9 # 1 # 1 1 # 1 2 # 1 3 # 1 4 # 1 5 # 1 6 # 1 7 # 1 8 # 1 9 # 2 # 2 1 SamKnows Probes 10 10

1

10

2

10

3

10

4

TCP connect time (msecs) 300 msec

YouTube Audio Stream

IPv4

# 1 # 2 # 3 # 4 # 5 # 6 # 7 # 9 # 1 # 1 1 # 1 2 # 1 3 # 1 4 # 1 5 # 1 6 # 1 7 # 1 8 # 1 9 # 2 # 2 1 SamKnows Probes 300 msec

YouTube Audio Stream

IPv6

# 1 # 2 # 3 # 4 # 5 # 6 # 7 # 8 # 9 # 1 # 1 1 # 1 2 # 1 3 # 1 4 # 1 5 # 1 6 # 1 7 # 1 8 # 1 9 # 2 # 2 1 SamKnows Probes 10 10

1

10

2

10

3

10

4

TCP connect time (msecs) 300 msec

YouTube Video Stream

IPv4

# 1 # 2 # 3 # 4 # 5 # 6 # 7 # 9 # 1 # 1 1 # 1 2 # 1 3 # 1 4 # 1 5 # 1 6 # 1 7 # 1 8 # 1 9 # 2 # 2 1 SamKnows Probes 300 msec

YouTube Video Stream

IPv6

15 / 25

slide-16
SLIDE 16

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | Happy Eyeballs

10 10

1

10

2

10

3

10

4

TCP connect time (msec) 0.0 0.2 0.4 0.6 0.8 1.0 CDF 300 msec

YouTube Video Stream

IPv6 IPv4

10 10

1

10

2

10

3

10

4

TCP connect time (msec) 0.0 0.2 0.4 0.6 0.8 1.0 CDF 300 msec

YouTube Audio Stream

IPv6 IPv4

IPV6 PREFERENCE VIDEO AUDIO #01 100.0% 100.0% #02 100.0% 100.0% #03 99.75% 99.75% #04 100.0% 100.0% #05 92.79% 92.79% #06 100.0% 100.0% #07 100.0% 100.0% #08 00.00% 00.00% #09 100.0% 100.0% #10 100.0% 100.0% #11 100.0% 100.0% #12 100.0% 100.0% #13 100.0% 100.0% #14 100.0% 100.0% #15 100.0% 100.0% #16 99.39% 98.94% #17 98.01% 98.26% #18 100.0% 100.0% #19 99.52% 96.85% #20 100.0% 100.0% #21 100.0% 100.0% 16 / 25

slide-17
SLIDE 17

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | Achievable Tiroughput and Stall Events

IPv4 IPv6

  • 2

4 6 20 40

Throughput(Mbps) Stall Duration(s)

1 2 3 4 5 6 7 8 9 1 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 2 1 2 2 2 3 1 2 3 4 5 6 7 8 9 1 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 2 1 2 2 2 3

Hour

  • GOOGLE (AS15169)

SEABONE (AS6762) YOUTUBE (AS36040)

17 / 25

slide-18
SLIDE 18

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Data Analysis | Google Global Caches

CATEGORY IPV4 n(PROBES) IPV6 n(PROBES) COMHEM (AS39651) 01

  • ASAHI (AS4685)

01

  • JAZZNET (AS12715)

01

  • EDPNET (AS9031)

01

  • DTAG (AS3320)

02 DTAG (AS3320) 02 CONTENT BIGLOBE (AS2518) 01

  • CACHES

ROEDUNET (AS2614) 01 ROEDUNET (AS2614) 01 NORDUNET (AS2603) 01 NORDUNET (AS2603) 01 BSKYB (AS5607) 01 BSKYB (AS5607) 01 SEABONE (AS6762) 01

  • QSC (AS20676)

01 QSC (AS20676) 01 NG (AS48161) 01

  • GOOGLE (AS15169)

20 GOOGLE (AS15169) 19 CDN YOUTUBE (AS43515) 03

  • YOUTUBE (AS36040)

02

  • LEVEL3 (AS3356)

01

  • IXP
  • INTERLAN (AS39107)

01

18 / 25

slide-19
SLIDE 19

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Conclusion

  • 1. TCP connect times to YouTube makes Happy Eyeballs prefer IPv6.
  • 2. Lower throughput is achieved when streaming YouTube over IPv6.
  • 3. YouTube content caches over IPv6 are largely absent.

Tie entire dataset is publicly released: http://www.netlab.tkk.fj/tutkimus/rtc/PAM2015

19 / 25

slide-20
SLIDE 20

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix

20 / 25

slide-21
SLIDE 21

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix | Summary

MA SUCCESS RATE STALL RATE SPEEDTEST (Mbps) GGC IPV4 IPV6 IPV4 IPV6 IPV4 IPV6 #01 100% 55% 0% 0% 92.56 72.35

  • #02

100% 100% 7% 1% 11.55 11.37

  • #03

100% 60% 0% 0% 61.82 57.99 IPV4 #04 100% 92% 0% 4% 10.68 7.55 IPV4 #05 100% 100% 29% 39% 1.49 1.47 IPv4 #06 100% 100% 0% 1% 27.83 6.16 IPv4 #07 100% 100% 0% 2% 44.24 43.45 IPv4 #08 100% 0% 0% 0% 13.14 9.80 IPv4 #09 100% 100% 0% 0% 83.20 25.06

  • #10

100% 55% 0% 0% 92.29 88.54

  • #11

100% 100% 0% 0% 37.87 39.10 BOTH #12 100% 91% 0% 0% 92.15 77.40

  • #13

100% 61% 0% 0% 217.99 170.46

  • #14

100% 99% 0% 0% 87.09 86.34 BOTH #15 96% 100% 0% 0% 10.99 10.82 BOTH #16 100% 100% 5% 30% 4.35 4.31 IPV4 #17 100% 100% 1% 57% 9.17 3.49

  • #18

100% 100% 0% 100% 20.80 0.29

  • #19

100% 99% 7% 5% 11.83 24.14

  • #20

100% 100% 0% 0% 93.37 91.83 BOTH #21 100% 100% 0% 0% 88.08 64.04

  • 21 / 25
slide-22
SLIDE 22

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix | Related Work

▶ YouTube Characterization:

▶ Gill et al. [5] (2007) study YouTube workload patterns in a campus. ▶ Cha et al. [6] (2007) study YouTube content popularity.

▶ Passive Measurements:

▶ Adhikari et al. [7] (2010) use fmow data to study YouTube from a tier-1 ISP. ▶ Finamore et al. [8] (2011) compare YouTube for mobile & PC devices. ▶ Dimopoulos et al. [9] (2013) study YouTube video sessions.

▶ Active Measurements:

▶ Juluri et al. [10] (2011) show Pytomo, a python tool that models a YouTube client. ▶ Adhikari et al. [11] (2012) use PlanetLab to crawl YouTube video ID space. ▶ Juluri et al. [12] (2013) use Pytomo to measure YouTube from 3 ISPs. ▶ Nam et al. [13] (2014) show YouSlow, browser plugin to detect live bufger stalls. 22 / 25

slide-23
SLIDE 23

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix | References I

[1] N. Sarrar, G. Maier, B. Ager, R. Sommer, and S. Uhlig, “Investigating IPv6 Traffjc,” in Passive and Active Measurement, ser. Lecture Notes in Computer Science, N. Tafu and F. Ricciato, Eds. Springer Berlin Heidelberg, 2012, vol. 7192, pp. 11–20. [Online]. Available: http://dx.doi.org/10.1007/978-3-642-28537-0_2 [2] D. Wing and A. Yourtchenko, “Happy Eyeballs: Success with Dual-Stack Hosts,” RFC 6555 (Proposed Standard), Internet Engineering Task Force, Apr. 2012. [Online]. Available: http://www.ietf.org/rfc/rfc6555.txt [3] S. Sundaresan, W. de Donato, N. Feamster, R. Teixeira, S. Crawford, and A. Pescapè, “Broadband internet performance: A view from the gateway,” ser. SIGCOMM ’11. ACM, 2011. [4] T. Stockhammer, “Dynamic adaptive streaming over http –: Standards and design principles,” in Proceedings of the Second Annual ACM Conference on Multimedia Systems, ser. MMSys ’11. New York, NY, USA: ACM, 2011,

  • pp. 133–144. [Online]. Available: http://dx.doi.org/10.1145/1943552.1943572

[5] P. Gill, M. Arlitt, Z. Li, and A. Mahanti, “Youtube Traffjc Characterization: A View from the Edge,” in Proceedings of the 7th ACM SIGCOMM Conference on Internet Measurement, ser. IMC ’07. New York, NY, USA: ACM, 2007, pp. 15–28. [Online]. Available: http://dx.doi.org/10.1145/1298306.1298310 [6] M. Cha, H. Kwak, P. Rodriguez, Y.-Y. Ahn, and S. Moon, “I Tube, You Tube, Everybody Tubes: Analyzing the World’s Largest User Generated Content Video System,” in Proceedings of the 7th ACM SIGCOMM Conference

  • n Internet Measurement, ser. IMC ’07.

New York, NY, USA: ACM, 2007, pp. 1–14. [Online]. Available: http://dx.doi.org/10.1145/1298306.1298309

23 / 25

slide-24
SLIDE 24

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix | References II

[7] V. K. Adhikari, S. Jain, and Z.-L. Zhang, “YouTube Traffjc Dynamics and Its Interplay with a Tier-1 ISP: An ISP Perspective,” in Proceedings of the 10th ACM SIGCOMM Conference on Internet Measurement, ser. IMC ’10. New York, NY, USA: ACM, 2010. [Online]. Available: http://dx.doi.org/10.1145/1879141.1879197 [8] A. Finamore, M. Mellia, M. M. Munafò, R. Torres, and S. G. Rao, “YouTube Everywhere: Impact of Device and Infrastructure Synergies on User Experience,” in Proceedings of the 2011 ACM SIGCOMM Conference on Internet Measurement Conference, ser. IMC ’11. New York, NY, USA: ACM, 2011, pp. 345–360. [Online]. Available: http://dx.doi.org/10.1145/2068816.2068849 [9] G. Dimopoulos, P. Barlet-Ros, and J. Sanjuas-Cuxart, “Analysis of youtube user experience from passive measurements,” in Network and Service Management (CNSM), 2013 9th International Conference on, Oct 2013,

  • pp. 260–267. [Online]. Available: http://dx.doi.org/10.1109/CNSM.2013.6727845

[10] P. Juluri, L. Plissonneau, and D. Medhi, “Pytomo: A tool for analyzing playback quality of youtube videos,” in Teletraffjc Congress (ITC), 2011, Sept 2011. [11] V. Adhikari, S. Jain, Y. Chen, and Z.-L. Zhang, “Vivisecting youtube: An active measurement study,” in INFOCOM, 2012 Proceedings IEEE, March 2012. [Online]. Available: http://dx.doi.org/10.1109/INFCOM.2012.6195644 [12] P. Juluri, L. Plissonneau, Y. Zeng, and D. Medhi, “Viewing youtube from a metropolitan area: What do users accessing from residential isps experience?” in IFIP/IEEE International Symposium on Integrated Network Management (IM),, May 2013.

24 / 25

slide-25
SLIDE 25

Measuring YouTube from Dual-Stacked Hosts Introduction

Motivation Research Question Research Contributions

Methodology

Metrics Measurement Setup Measurement Setup Measurement Trials

Data Analysis

Speed Tests TCP Connect Times Happy Eyeballs Tiroughput and Stall Events Google Global Caches

Conclusion Appendix

Appendix | References III

[13] H. Nam, K.-H. Kim, D. Calin, and H. Schulzrinne, “Youslow: a performance analysis tool for adaptive bitrate video streaming,” in Proceedings of the 2014 ACM conference on SIGCOMM. ACM, 2014, pp. 111–112.

25 / 25