26
Peer-to-Peer Introduction to a New Set of Opportunities Gwendal Simon Department of Computer Science TELECOM Bretagne Jan. 2008

peer-to-peer oppotunities

Embed Size (px)

DESCRIPTION

A short tour about peer-to-peer applications, and their opportunities, in Jan 2008. Attendees were members of a "research and development cluster" on multimedia and networking

Citation preview

Page 1: peer-to-peer oppotunities

Peer-to-PeerIntroduction to a New Set of Opportunities

Gwendal Simon

Department of Computer ScienceTELECOM Bretagne

Jan. 2008

Page 2: peer-to-peer oppotunities

2/ 17 Birth

Context : Napster Shutdown• early adopters are addicted• early majority is just joining• legal perspectives are dark

Server

0 1 2 3 4 5

Gnutella : Serverless Napster• simple but bad idea• just a piece of software• viral bazaar diffusion

0001

02

03 04

05

06 07

08

09

10

11 12

13

14

15

16

17

18

19

Gwendal Simon : Peer-to-Peer

Page 3: peer-to-peer oppotunities

2/ 17 Birth

Context : Napster Shutdown• early adopters are addicted• early majority is just joining• legal perspectives are dark

Server

0 1 2 3 4 5

Gnutella : Serverless Napster• simple but bad idea• just a piece of software• viral bazaar diffusion

0001

02

03 04

05

06 07

08

09

10

11 12

13

14

15

16

17

18

19

Gwendal Simon : Peer-to-Peer

Page 4: peer-to-peer oppotunities

2/ 17 Birth

Context : Napster Shutdown• early adopters are addicted• early majority is just joining• legal perspectives are dark

Server

0 1 2 3 4 5

Gnutella : Serverless Napster• simple but bad idea• just a piece of software• viral bazaar diffusion

0001

02

03 04

05

06 07

08

09

10

11 12

13

14

15

16

17

18

19

Gwendal Simon : Peer-to-Peer

Page 5: peer-to-peer oppotunities

3/ 17

Gwendal Simon : Peer-to-Peer

Page 6: peer-to-peer oppotunities

4/ 17 Peer-to-Peer is not only File Sharing

. . .but also large-scale legal applications :• where peers just have to be introduced : phone, storage. . .• where server become bottleneck : broadcasting . . .• where nobody should be responsible : virtual world, forums. . .

Attractive Issues : no governance, nor understanding• academic : what exactly is scalable ?• developers : is the Bazaar consistent ?• users : what is the value when I contribute ?

Gwendal Simon : Peer-to-Peer

Page 7: peer-to-peer oppotunities

4/ 17 Peer-to-Peer is not only File Sharing

. . .but also large-scale legal applications :• where peers just have to be introduced : phone, storage. . .• where server become bottleneck : broadcasting . . .• where nobody should be responsible : virtual world, forums. . .

Attractive Issues : no governance, nor understanding• academic : what exactly is scalable ?• developers : is the Bazaar consistent ?• users : what is the value when I contribute ?

Gwendal Simon : Peer-to-Peer

Page 8: peer-to-peer oppotunities

5/ 17

Part I

Is there any Opportunity in the Bazaar ?

Gwendal Simon : Peer-to-Peer

Page 9: peer-to-peer oppotunities

6/ 17 Motivations

network empirical law value per user global valuebroadcast (TV) Sarnoff 1 none-to-one (phone) Metcalfe n n2

peer-to-peer Reed 2n 2n

Stirring up a community is the actual challenge :• aggregate a vast population• incite them to contribute

⇒ joining an existing community may be far easier(1047 projects referenced in Sourceforge)

Gwendal Simon : Peer-to-Peer

Page 10: peer-to-peer oppotunities

6/ 17 Motivations

network empirical law value per user global valuebroadcast (TV) Sarnoff 1 none-to-one (phone) Metcalfe n n2

peer-to-peer Reed 2n 2n

Stirring up a community is the actual challenge :• aggregate a vast population• incite them to contribute

⇒ joining an existing community may be far easier(1047 projects referenced in Sourceforge)

Gwendal Simon : Peer-to-Peer

Page 11: peer-to-peer oppotunities

7/ 17 Yahoo’s Gnutella Developer Forum

"a forum for active Gnutella developers without distractions"• developers from a company hosting an "official" client• independent open-source developers• just some curious guys. . .

Emerging usage : normalization forum• protocol update, feature proposal, idea exchanges• Collateral usage : innovation center

• feedback from users : chunk swarming• external opportunities : URN by Bitzi• technical development : bootstrap, NAT traversal

Gwendal Simon : Peer-to-Peer

Page 12: peer-to-peer oppotunities

8/ 17 Innovator’s Dilemma

Cooperative or competitive behavior ?• actors are competitors : have to differentiate• the value is in Gnutella usefulness : have to cooperate

Conciliate gain for the user and progressive adoption• no innovation should require full agreement• for most innovations, gain grows with the number of adopters• individual innovations (e.g. ergonomics) allow differentiating

Gwendal Simon : Peer-to-Peer

Page 13: peer-to-peer oppotunities

9/ 17 Business Opportunity

A complex ecosystem :• leery of market economy :

• bazaar culture implies free software• contributors are reluctant to pay

• but attractive :• several millions of active users• a technophile population of online influencers

Requires a hook mixing technical expertise and social understanding• out of the peer-to-peer system• but in the peer-to-peer activities

Gwendal Simon : Peer-to-Peer

Page 14: peer-to-peer oppotunities

10/ 17

Part II

When Academic Studies Become Reality

Gwendal Simon : Peer-to-Peer

Page 15: peer-to-peer oppotunities

11/ 17 Peer-to-Peer Fundamentals

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

Gwendal Simon : Peer-to-Peer

Page 16: peer-to-peer oppotunities

11/ 17 Peer-to-Peer Fundamentals

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

Gwendal Simon : Peer-to-Peer

Page 17: peer-to-peer oppotunities

11/ 17 Peer-to-Peer Fundamentals

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

00

01

02

03

04

05

06

07

08

09

10

11

12

13

14

15

16

17

18

19

20

21

22

2324

25

2627

28

29

30

31

→ each peer chooses some direct neighbors among all peers=⇒ it creates an overlay

Gwendal Simon : Peer-to-Peer

Page 18: peer-to-peer oppotunities

12/ 17 Few Mature Works

Scientists have focused on few services :• diffusion of large files (bit-torrent)• distributed hash tables (DHT) :

• store a pair (key, resource) in the system• give the key, retrieve the related resource• if a resource has been stored, it must be accessible

Attractive implementations for controlled network :• server farm : managing a vast amount of resources• content delivery network (CDN) : pushing contents• network of boxes : the new eldorado

Gwendal Simon : Peer-to-Peer

Page 19: peer-to-peer oppotunities

13/ 17 DHT Principles (Chord)

000102

04

0607

0809

1012131416181920

2123

25

28

3031

3233

37

39

41

4447 50

57

63

i + 23i + 24

i + 25

16 → 50

50 → 02

02 → 06

Gwendal Simon : Peer-to-Peer

Page 20: peer-to-peer oppotunities

13/ 17 DHT Principles (Chord)

000102

04

0607

0809

1012131416181920

2123

25

28

3031

3233

37

39

41

4447 50

57

63

i + 23i + 24

i + 25

16 → 50

50 → 02

02 → 06

Gwendal Simon : Peer-to-Peer

Page 21: peer-to-peer oppotunities

13/ 17 DHT Principles (Chord)

000102

04

0607

0809

1012131416181920

2123

25

28

3031

3233

37

39

41

4447 50

57

63

i + 23i + 24

i + 25

16 → 50

50 → 02

02 → 06

Gwendal Simon : Peer-to-Peer

Page 22: peer-to-peer oppotunities

14/ 17 Business opportunities

Any cost reduction on existing services ?• one overlay per need : initial development may be costly• decreasing server cost : gain for large-scale systems

But new services are possible :• lightweight and immediately worldwide• relying on networking skills

Gwendal Simon : Peer-to-Peer

Page 23: peer-to-peer oppotunities

15/ 17

Part III

Conclusive Thoughts

Gwendal Simon : Peer-to-Peer

Page 24: peer-to-peer oppotunities

16/ 17 I&R activities on peer-to-peer

Few projects claim studying peer-to-peer :• Solipsis : a bazaar peer-to-peer virtual world• P2P-images : a industrial peer-to-peer streaming system

However many academic skills on peer-to-peer in Bretagne :• computer sciences : world-class teams at IRISA and Telecom Bretagne• social and human sciences : M@rsouin gathers high-level institutes

Gwendal Simon : Peer-to-Peer

Page 25: peer-to-peer oppotunities

17/ 17 A ShiftAlong with Web2.0 trend :

• capturing money requires complex models• empower a community → beta web-based eye-candy• both technical and social wow effect

Also a deep scientific trend :• toward “one billion computers per human”• balancing power between edges and core network

Few possible projects :• self-administered network : distributed package management• mixed reality : distributed “1.5 life”• tangible gifts : distributed Wikipedia

Gwendal Simon : Peer-to-Peer

Page 26: peer-to-peer oppotunities

17/ 17 A ShiftAlong with Web2.0 trend :

• capturing money requires complex models• empower a community → beta web-based eye-candy• both technical and social wow effect

Also a deep scientific trend :• toward “one billion computers per human”• balancing power between edges and core network

Few possible projects :• self-administered network : distributed package management• mixed reality : distributed “1.5 life”• tangible gifts : distributed Wikipedia

Gwendal Simon : Peer-to-Peer