Poor Video Streaming Performance Explained (and Fixed)

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

HTTP-based video streaming is a key application on the Internet today, comprising the majority of Internet traffic. Yet customers remain dissatisfied with video quality, resulting in lost revenue for content providers. Recent studies have blamed this on the adaptive bitrate selection (ABR) algorithm used by client players, claiming it interacts poorly with TCP when the video buffer is full, which causes it to underestimate available network bandwidth.We show that the root cause of the problem lies in the data plane, and that even a perfect control plane (ABR) algorithm is not enough to guarantee video flows their fair share of network bandwidth. Namely, it is the sequential download of video segments that is at fault, as they disrupt the normal interaction between TCP congestion control and router queue occupancy. We carefully model and characterize the behavior of streaming video according to download size and network conditions, and use this to develop an adaptive algorithm for optimally controlling download behavior. Our approach achieves near-optimal throughput and fast bitrate adaptation, regardless of the control plane algorithm.We implement our approach as a DASH video player called Sprint, and evaluate it against state-of-the-art proposals from the literature as well as deployed players from Netflix, YouTube, Hulu, and Amazon. Sprint consistently achieves above 90% of its fair-share throughput, while the previous state-of-the-art exhibits high variability (e.g., 31% under some network conditions, close to fair share in others). Our measurements show industry players often achieve below 50% of their fair share.

Original languageEnglish (US)
Title of host publicationApplicative 2016
PublisherAssociation for Computing Machinery, Inc
ISBN (Print)9781450344647
DOIs
StatePublished - Jun 1 2016
EventApplicative Conference 2016 - New York, United States
Duration: Jun 1 2016Jun 2 2016

Publication series

NameApplicative 2016

Conference

ConferenceApplicative Conference 2016
CountryUnited States
CityNew York
Period6/1/166/2/16

Fingerprint

Video streaming
Throughput
Internet
Bandwidth
HTTP
Adaptive algorithms
Routers
Industry

All Science Journal Classification (ASJC) codes

  • Computer Science(all)

Cite this

Freedman, M. J. (2016). Poor Video Streaming Performance Explained (and Fixed). In Applicative 2016 (Applicative 2016). Association for Computing Machinery, Inc. https://doi.org/10.1145/2959689.2960158
Freedman, Michael Joseph. / Poor Video Streaming Performance Explained (and Fixed). Applicative 2016. Association for Computing Machinery, Inc, 2016. (Applicative 2016).
@inproceedings{28058fc20d754d7190e13c16b62ff81d,
title = "Poor Video Streaming Performance Explained (and Fixed)",
abstract = "HTTP-based video streaming is a key application on the Internet today, comprising the majority of Internet traffic. Yet customers remain dissatisfied with video quality, resulting in lost revenue for content providers. Recent studies have blamed this on the adaptive bitrate selection (ABR) algorithm used by client players, claiming it interacts poorly with TCP when the video buffer is full, which causes it to underestimate available network bandwidth.We show that the root cause of the problem lies in the data plane, and that even a perfect control plane (ABR) algorithm is not enough to guarantee video flows their fair share of network bandwidth. Namely, it is the sequential download of video segments that is at fault, as they disrupt the normal interaction between TCP congestion control and router queue occupancy. We carefully model and characterize the behavior of streaming video according to download size and network conditions, and use this to develop an adaptive algorithm for optimally controlling download behavior. Our approach achieves near-optimal throughput and fast bitrate adaptation, regardless of the control plane algorithm.We implement our approach as a DASH video player called Sprint, and evaluate it against state-of-the-art proposals from the literature as well as deployed players from Netflix, YouTube, Hulu, and Amazon. Sprint consistently achieves above 90{\%} of its fair-share throughput, while the previous state-of-the-art exhibits high variability (e.g., 31{\%} under some network conditions, close to fair share in others). Our measurements show industry players often achieve below 50{\%} of their fair share.",
author = "Freedman, {Michael Joseph}",
year = "2016",
month = "6",
day = "1",
doi = "10.1145/2959689.2960158",
language = "English (US)",
isbn = "9781450344647",
series = "Applicative 2016",
publisher = "Association for Computing Machinery, Inc",
booktitle = "Applicative 2016",

}

Freedman, MJ 2016, Poor Video Streaming Performance Explained (and Fixed). in Applicative 2016. Applicative 2016, Association for Computing Machinery, Inc, Applicative Conference 2016, New York, United States, 6/1/16. https://doi.org/10.1145/2959689.2960158

Poor Video Streaming Performance Explained (and Fixed). / Freedman, Michael Joseph.

Applicative 2016. Association for Computing Machinery, Inc, 2016. (Applicative 2016).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Poor Video Streaming Performance Explained (and Fixed)

AU - Freedman, Michael Joseph

PY - 2016/6/1

Y1 - 2016/6/1

N2 - HTTP-based video streaming is a key application on the Internet today, comprising the majority of Internet traffic. Yet customers remain dissatisfied with video quality, resulting in lost revenue for content providers. Recent studies have blamed this on the adaptive bitrate selection (ABR) algorithm used by client players, claiming it interacts poorly with TCP when the video buffer is full, which causes it to underestimate available network bandwidth.We show that the root cause of the problem lies in the data plane, and that even a perfect control plane (ABR) algorithm is not enough to guarantee video flows their fair share of network bandwidth. Namely, it is the sequential download of video segments that is at fault, as they disrupt the normal interaction between TCP congestion control and router queue occupancy. We carefully model and characterize the behavior of streaming video according to download size and network conditions, and use this to develop an adaptive algorithm for optimally controlling download behavior. Our approach achieves near-optimal throughput and fast bitrate adaptation, regardless of the control plane algorithm.We implement our approach as a DASH video player called Sprint, and evaluate it against state-of-the-art proposals from the literature as well as deployed players from Netflix, YouTube, Hulu, and Amazon. Sprint consistently achieves above 90% of its fair-share throughput, while the previous state-of-the-art exhibits high variability (e.g., 31% under some network conditions, close to fair share in others). Our measurements show industry players often achieve below 50% of their fair share.

AB - HTTP-based video streaming is a key application on the Internet today, comprising the majority of Internet traffic. Yet customers remain dissatisfied with video quality, resulting in lost revenue for content providers. Recent studies have blamed this on the adaptive bitrate selection (ABR) algorithm used by client players, claiming it interacts poorly with TCP when the video buffer is full, which causes it to underestimate available network bandwidth.We show that the root cause of the problem lies in the data plane, and that even a perfect control plane (ABR) algorithm is not enough to guarantee video flows their fair share of network bandwidth. Namely, it is the sequential download of video segments that is at fault, as they disrupt the normal interaction between TCP congestion control and router queue occupancy. We carefully model and characterize the behavior of streaming video according to download size and network conditions, and use this to develop an adaptive algorithm for optimally controlling download behavior. Our approach achieves near-optimal throughput and fast bitrate adaptation, regardless of the control plane algorithm.We implement our approach as a DASH video player called Sprint, and evaluate it against state-of-the-art proposals from the literature as well as deployed players from Netflix, YouTube, Hulu, and Amazon. Sprint consistently achieves above 90% of its fair-share throughput, while the previous state-of-the-art exhibits high variability (e.g., 31% under some network conditions, close to fair share in others). Our measurements show industry players often achieve below 50% of their fair share.

UR - http://www.scopus.com/inward/record.url?scp=85055356432&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055356432&partnerID=8YFLogxK

U2 - 10.1145/2959689.2960158

DO - 10.1145/2959689.2960158

M3 - Conference contribution

SN - 9781450344647

T3 - Applicative 2016

BT - Applicative 2016

PB - Association for Computing Machinery, Inc

ER -

Freedman MJ. Poor Video Streaming Performance Explained (and Fixed). In Applicative 2016. Association for Computing Machinery, Inc. 2016. (Applicative 2016). https://doi.org/10.1145/2959689.2960158