Reducing BitTorrent Traffic at the Internet Scale Stevens Le Blond , - - PowerPoint PPT Presentation

▶

Jul 12, 2023 213 likes •368 views

Reducing BitTorrent Traffic at the Internet Scale Stevens Le Blond , Arnaud Legout, Walid Dabbous 1 Two open questions about BitTorrent locality AS 2 AS 3 # connections Internet Internet AS 1 Reduces traffic AS 4 AS 5 How far can we

SLIDE 1

1

Reducing BitTorrent Traffic at the Internet Scale

Stevens Le Blond, Arnaud Legout, Walid Dabbous

SLIDE 2

2

Two open questions about BitTorrent locality

AS 1 # connections   How far can we push BitTorrent locality?   How much inter-AS traffic reduction?   How much performance penalty?  How much savings at the Internet scale?   Do we gain significantly for big torrents?   Do small torrents (with no locality gain) dominate? AS 2 AS 3 Internet Internet AS 4 AS 5 Reduces traffic

SLIDE 3

3

Roadmap

 Methodology   How far can we push How far can we push BitTorrent BitTorrent locality? locality?   How much savings at the Internet Scale? How much savings at the Internet Scale?

SLIDE 4

4

Methodology

Experimental setup  Real BitTorrent clients

Partition Merging (PM)

 Tracker

assigns clients to virtual ASes and

control number of inter-AS connections  Up to 10,000 peers  20kB/s upload, 100MB file, 256kB pieces

SLIDE 5

5

Methodology

Parameters  Homogeneous torrents  Heterogeneous torrents

Real AS distribution

Metrics  Overhead  Slowdown  95-th percentile

SLIDE 6

6

Roadmap

  Methodology Methodology  How far can we push BitTorrent locality?   How much savings at the Internet scale? How much savings at the Internet scale?

SLIDE 7

7

Real distribution of peers per AS

790,717 .torrent files 214,443 active torrents with 6,113,224 peers Crawler … Trackers

SLIDE 8

8

Distribution of peers in the reference torrent

# ASes (log scale) # peers (log scale) 9,844 peers 1,043 ASes 386 peers in biggest AS 458 ASes with 1 peer

SLIDE 9

9 # of peers per AS (log scale) Overhead (log sclae)

Impact of real distributions of peers per AS

n overhead

SLIDE 10

10

Impact of real distributions of peers per AS

n slowdown

# of peers per AS (log scale) Slowdown

Big ASes get partitioned No partitions

Orders of magnitude less overhead without performance penalty

SLIDE 11

11

Roadmap

  Methodology Methodology   How far can we push How far can we push BitTorrent BitTorrent locality? locality?  How much savings at the Internet scale?

SLIDE 12

12

How to find overhead savings for all torrents?

 Experiment with multiple reference torrents

English-language Movie
Italian-language Movie
Games and small torrents

 Overhead mainly depends on # of peers per AS  Compute overhead for all 214,443 torrents

SLIDE 13

13

Inter-AS traffic at the Internet scale

Cumulated inter-AS traffic (TB) Torrent ID (sorted by decreasing inter-AS traffic, log scale)

Large savings at the Internet scale

11.6PB 41% savings 56% savings

SLIDE 14

14

Take home messages

  Orders of magnitude less overhead for ISPs without performance penalty for end-users 2) Large savings at the Internet scale, i.e., 41% savings on 11.6 petabytes of inter-AS traffic Contact: stevens.le_blond@inria.fr