Measuring YouTube Content Delivery over IPv6 Q/A Recommendations - PowerPoint PPT Presentation

Motivation Prague, CZ July 2017 Jacobs University Bremen, Germany Jürgen Schönwälder TU Munich, Germany Jörg Ott Aalto University, Helsinki, Finland Saba Ahsan Joint work with SIGCOMM Computer Communication Review Published at: IETF 99 Methodology TU Munich Vaibhav Bajpai Measuring YouTube Content Delivery over IPv6 Q/A Recommendations Stall Events Tiroughput Startup Delay TCP connect times IPv6 Preference Success Rate 1 / 18 July 2017: https://goo.gl/oUJ7Ej

Motivation Recommendations Methodology Do users experience benefjt (or sufger) from YouTube streaming over IPv6? shaded region represents the duration of the longitudinal study. Motivation Q/A 2 / 18 Stall Events Success Rate Startup Delay TCP connect times IPv6 Preference Tiroughput Google IPv6 Adoption 20% ▶ IPv6 contributes ∼ 25% [1] of traffjc within Comcast. 15% 10% ▶ Swisscom reports ∼ 60% [1] of IPv6 traffjc is YouTube. 5% 0% ▶ IPv6 traffjc largely dominated by YouTube [2]. 2009 2010 2011 2012 2013 2014 2015 2016 2017 NETWORK TYPE # RESIDENTIAL 78 NREN / RESEARCH 10 BUSINESS / DATACENTER 08 OPERATOR LAB 04 IXP 01 RIR # RIPE 60 ARIN 29 APNIC 10 AFRINIC 01 LACNIC 01 ∼ 100 dual-stacked SamKnows probes ( ∼ 66 difgerent origin ASes)

Motivation Stall Events Tiis is the fjrst study to measure YouTube content delivery over IPv6 Research Contribution Methodology Recommendations Q/A Tiroughput Startup Delay TCP connect times IPv6 Preference Success Rate 3 / 18 ▶ HE (RFC 6555) makes clients to prefer streaming YouTube videos over IPv6. ▶ Observed performance (both in terms of latency and throughput) over IPv6 is worse. ▶ Stall rates are low, bitrates that can be reliably streamed are comparable. ▶ When a stall occurs, stall durations over IPv6 are higher. ▶ Worse performance is due to GGC nodes that are IPv4-only.

Motivation Recommendations 3. No regional restrictions. 2. Available in Full HD. 1. Video duration > 60s. Methodology Q/A Methodology | Selection of YouTube Videos Stall Events Tiroughput Startup Delay TCP connect times IPv6 Preference Success Rate 4 / 18 ▶ Using YouTube v3 API [3]. ▶ Video Selection Criteria: ▶ List updated every 12h. ▶ Probes daily pull the list. ▶ Tie test supports non-adaptive and step-down playout modes only. ▶ Results are biased our vantage points (centered largely around EU, US and JP).

Motivation Recommendations Methodology hour (over both AF). iterations to total iterations. Success Rate Q/A 5 / 18 Stall Events Success Rate Startup Delay Tiroughput TCP connect times IPv6 Preference Success Rate 1.0 0.8 IPv6 (102) 0.6 ▶ Number of successful 0.4 ['14 - '17] 0.2 0.0 CCDF 1.0 ▶ Tie test executes once every 0.8 IPv4 (102) 0.6 0.4 0.2 0.0 80% 85% 90% 95% 100% ▶ 99% of probes achieve success rate of more than 94% over IPv4. ▶ 97% of probes achieve success rate of more than 94% over IPv6. ▶ Slightly lower success rates over IPv6 due to network issues closer to probes.

Motivation Recommendations Clients prefer streaming YouTube videos over IPv6 Methodology IPv6 Preference Q/A 6 / 18 Stall Events Tiroughput Startup Delay TCP connect times Success Rate IPv6 Preference IPv6 Preference 1.0 ['14 - '17] 0.8 Web (871) 0.6 CCDF Audio (871) 0.4 Video (871) 0.2 0.0 95% 96% 97% 98% 99% 100% ▶ RFC 6724 [4] makes apps prefer connections made over IPv6. ▶ RFC 6555 [5] allows apps to fallback to IPv4 when IPv6 connectivity is bad. ▶ TCP connections over IPv6 are preferred at least 97% of the time.

Motivation Stall Events Methodology TCP connect times Q/A Recommendations 7 / 18 Tiroughput TCP connect times IPv6 Preference Success Rate Startup Delay TCP connect times 1.0 ['14 - '17] Web (6.5M) 0.8 Audio (6.5M) 0.6 CDF 0.4 Video (6.5M) 0.2 0.0 −100 −75 −50 −25 0 25 50 ∆t (ms) ▶ 63% of a/v streams (and 72% of the web connections) are slower over IPv6. ▶ 14% of a/v streams are at least 10 ms slower over IPv6.

Motivation Stall Events Methodology TCP connect times Q/A Recommendations 8 / 18 Tiroughput Startup Delay TCP connect times IPv6 Preference Success Rate TCP Connect Times 0 ∆t (ms) Web −5 TCP Connect Times 0.0 ∆t (ms) −0.1 Audio −0.2 Video −0.3 −0.4 Jan Jul Jan Jul Jan Jul 2015 2016 2017 ▶ TCP connect times consistently higher over IPv6 and have not improved over time. ▶ TCP connect times towards the webpage worse over IPv6 than towards media servers.

Motivation Methodology Success Rate IPv6 Preference TCP connect times Startup Delay Tiroughput Stall Events Recommendations Q/A Sequence Diagram (contd.) 9 / 18

Motivation Stall Events Methodology Startup Delay Q/A Recommendations 10 / 18 Tiroughput TCP connect times IPv6 Preference Success Rate Startup Delay Startup Delay 1.0 ['14 - '17] 6.5M 0.8 0.6 CDF 0.4 0.2 0.0 −5 −4 −3 −2 −1 0 1 ∆s (s) ▶ 80% of the samples are slower over IPv6. ▶ Half of the samples are at least 100 ms slower over IPv6.

Motivation Methodology Success Rate IPv6 Preference TCP connect times Startup Delay Tiroughput Stall Events Recommendations Q/A Startup Delay 11 / 18 ▶ Prebufgering durations are ∼ 25 ms higher over IPv6. ▶ Startup delays are ∼ 100 ms higher over IPv6. ▶ Initial interaction with the web server makes startup delay worse over IPv6.

Motivation Methodology Success Rate IPv6 Preference TCP connect times Startup Delay Tiroughput Stall Events Recommendations Q/A Sequence Diagram (contd.) 12 / 18

Motivation Stall Events Methodology Tiroughput Q/A Recommendations 13 / 18 Tiroughput TCP connect times IPv6 Preference Startup Delay Success Rate Throughput 1.0 ['14 - '17] Video (6.5M) 0.8 Audio (6.5M) 0.6 CDF 0.4 0.2 0.0 −4 −2 0 2 4 ∆tp (MB/s) Throughput 0.2 ∆tp (MB/s) −0.1 −0.4 −0.7 Video Audio −1.0 Jul Jul Jan Jan Jan Jul 2015 2016 2017 ▶ 80% of video and 60% audio samples achieve lower throughput over IPv6. ▶ Tie throughput is consistently lower over IPv6, but it has improved over time.

Motivation Methodology Success Rate IPv6 Preference TCP connect times Startup Delay Tiroughput Stall Events Recommendations Q/A Sequence Diagram (contd.) 14 / 18

Motivation Stall Events Methodology Stall Rates Q/A Recommendations 15 / 18 Tiroughput TCP connect times IPv6 Preference Startup Delay Success Rate Stall Rate 1.0 0.8 0.6 0.4 IPv6 (102) ['14 - '17] 0.2 0.0 CDF 1.0 0.8 0.6 0.4 IPv4 (102) 0.2 0.0 0% 5% 10% 15% 20% 25% 30% ▶ 90% of the probes witness less than 1% stall rate over both address families. ▶ Bitrates reliably streamed is also comparable over both address families.

Motivation Stall Events Methodology Stall Durations Q/A Recommendations 16 / 18 Tiroughput TCP connect times IPv6 Preference Success Rate Startup Delay Stall Durations 1.0 ['14 - '17] 0.8 0.6 CDF 0.4 1.7K 0.2 0.0 −60 −40 −20 0 20 40 60 ∆st (s) ▶ 80% of samples experience stall durations that are at least 1s longer.

Motivation Stall Events Methodology Recommendations Q/A Recommendations 17 / 18 Tiroughput TCP connect times IPv6 Preference Startup Delay Success Rate ▶ Update RFC 6555 with a lower HE timer value. ▶ We have shown [6] that reducing HE timer value to 150 ms (from 300 ms) helps. 6to4/Teredo Decline 0.20% 0.15% W6D W6LD 0.10% 0.05% 0.00% 2009 2010 2011 2012 2013 2014 2015 2016 2017 ▶ ISPs should put latency as a fjrst-class citizen. ▶ ISPs should ensure GGC nodes are dual-stacked. ▶ Request an IPv6 prefjx allocation from Google.

Motivation Recommendations bajpaiv@in.tum.de | @bajpaivaibhav www.vaibhavbajpai.com Methodology Q/A Takeway Stall Events Tiroughput Startup Delay TCP connect times IPv6 Preference Success Rate 18 / 18 ▶ Clients prefer streaming YouTube videos over IPv6. ▶ Observed performance (both in terms of latency and throughput) over IPv6 is worse. ▶ Stall rates are low, but when a stall occurs, stall durations over IPv6 are higher. ▶ Worse performance due to GGC nodes that are IPv4-only. ▶ Reproducibility Considerations: ▶ Tie test is open-sourced: https://github.com/sabyahsan/youtube-test ▶ Tie dataset is released: https://github.com/vbajpai/2017-ccr-youtube-analysis