confused, timid and unstable: picking a video rate is hard te-yuan (ty) huang stanford university...
TRANSCRIPT
![Page 1: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/1.jpg)
Confused, Timid and Unstable:Picking a Video Rate is Hard
Te-Yuan (TY) HuangStanford University
Nov 15th, 2012
Joint work with Nikhil Handigol, Brandon Heller, Nick McKeown, Ramesh Johari
![Page 2: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/2.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 2
Video is the BIG thing on the Internet
• Video is more than 50% of peak traffic in the US• Trend: Streaming over HTTP– Content Distribution Networks (CDNs)
• Well-provisioned HTTP servers at the edge of the Internet
• Cheap (2-3 cents per GB in 2011)
– Firewall friendliness
![Page 3: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/3.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 3
![Page 4: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/4.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 4
Before download started
Afterdownload started
![Page 5: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/5.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 5
What happened?
• Both the download and video are over HTTP– TCP shares my home link equally among all flows
What is the problem?
![Page 6: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/6.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 6
Experiment Setup
The Internet
Bandwidth Controller
CDN 1
CDN 2
CDN 3
Content Distribution Networks
![Page 7: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/7.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 7
Video Ratein the Presence of a Competing Flow
Video Rate
AvailableVideo Rates
Fair share
Competing FlowThroughout
Steps down all the way to the lowest quality
Video FlowThroughput
![Page 8: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/8.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 8
What If We Manually Select a Video Rate?
AvailableVideo Rates Manually select
1750kb/s
Video FlowThroughput Video keeps playing
without any problem
Video Rate
![Page 9: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/9.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 9
What If We Manually Select a Video Rate?Competing FlowThroughput
Video Rate
AvailableVideo Rates Manually select
1750kb/s
Competing flow throughput is brought downVideo Flow
Throughput
![Page 10: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/10.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 10
Not Just One Service’s Problem
• This happens in all the services we measured– Hulu, Netflix, Vudu
Video Rate
Video Rate
![Page 11: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/11.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 11
The Problem
Why?
It picks a video rate that is much too low
Video client ends up with much less throughput than its fair share
![Page 12: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/12.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 12
Outline
• How does rate selection over HTTP work?• The rate selection process• Where does it go wrong?• The complete story• Conclusion
![Page 13: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/13.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 13
How Does Rate Selection over HTTP Work?
The Internet
Client
Standard, commoditized HTTP Servers
Rate selection logic resides at client side
CDN 3
File 1:
File 2:
1750 1750
1050 1050 1050Get File 1 (1750kb/s)
Serve the video with quality 1750kb/s
PlayoutBuffer
![Page 14: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/14.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 14
The Rate Selection Process
Initialvideo rate
Video rate for the next video segment
BandwidthestimationDownload
& measure Pick a rate
What goes wrong?
![Page 15: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/15.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 15
The Rate Selection Process
BandwidthestimationDownload
& measure Pick a rate
Initialvideo rate
Video rate for the next video segment
![Page 16: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/16.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 16
TCP Throughput of the Video Flow
Playout Buffer is Full
TCP Throughput
OFF Period: 1-2 seconds
![Page 17: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/17.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 17
Impact of OFF period on TCP
• TCP sender resets its congestion window– When idle more than one RTO (200ms)– Slow-start restart, RFC 2581/5681– Linux 3.x (tcp_output.c, line 163)
• Throughput will be affected– Worse with a competing flow• Experience packet loss during slow start
50% of the segments get < 1.8Mb/s(Fair Share is 2.5Mb/s)
![Page 18: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/18.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 18
The Rate Selection Process
Bandwidthestimation
Download& measure Pick a rate
Underestimation
If perceived 1.8 Mb/s, which video rate would it pick?
1400 kb/s?1750 kb/s? Even lower?
Video rate for the next video segment
Initialvideo rate
![Page 19: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/19.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 19
Conservative Rate Selection
Available BW
Video Rate
19
If 1.8Mb/s BW is perceived,the client switches down to1050kb/s video rate
![Page 20: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/20.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 20
The Rate Selection Process
Bandwidthestimation
Download& measure
Pick a rateUnderestimation
Conservatively
Is there any consequence of being conservative?
Initialvideo rate
Video rate for the next video segment
![Page 21: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/21.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 21
Smaller Segment Size for Lower Video RateWhen requesting a smaller segment size, lower probability of obtaining fair share.
![Page 22: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/22.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 22
Lower video rate leads to further bandwidth underestimation
Lowest Video Rate
Highest Video Rate
![Page 23: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/23.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 23
The Rate Selection Process
Bandwidthestimation
Download& measure
Pick a rateUnderestimation
Conservatively
Request for a smaller segment
Further
Video rate for the next video segment
Initialvideo rate
![Page 24: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/24.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 24
The Complete Story
Video QualityBandwidth Underestimation
Conservatism
Further Underestimation
![Page 25: Confused, Timid and Unstable: Picking a Video Rate is Hard Te-Yuan (TY) Huang Stanford University Nov 15 th, 2012 Joint work with Nikhil Handigol, Brandon](https://reader038.vdocuments.site/reader038/viewer/2022103005/56649d1a5503460f949ef238/html5/thumbnails/25.jpg)
Te-Yuan (TY) Huang @ ACM IMC'12, Boston 25
Conclusion
• The problem: Video flow gets a much lower throughput than its fair share– Problem occurred in all the services we measured
• The reason: Video client tries to do TCP’s job – It is afraid of causing network congestion
• It becomes timid
– It tries to figure out its available bandwidth• Without seeing the interaction between the playout buffer, segment
sizes and TCP dynamics• It is easily confused and becomes unstable
• One service provider changed its algorithm because of this work