topology mapping

Topology Mapping

Bo Sheng

Sept. 15

Outline

Overview Solutions

LTM ACE

Problems and discussion Conclusion

Introduction

Topology mapping Mismatch between overlay and physical

infrastructure Topology optimization

Introduction

Traffic problem Facts

95% of any pairs of Gnutella nodes are within 7 hops 50,000 nodes generate 1G/second, 330T/month

Reasons Blind flooding

Cycles, merge of multiple paths, neighbors exchange Topology problem

Multiple times over a physical link

Introduction

Perfect match

S S

Network infrastructure Overlay network

Introduction

Mismatch

S S

N1

N2

N3

Network infrastructure Overlay network

23

4

5

52

4

Topology Mismatch

Problems Randomly choosing neighbors Logically close, but physically far away

S PN1 N2

Topology Mismatch

Problems Unnecessary traffic

Inefficient utilization of bandwidth Only 2%~5% Gnutella connections link nodes within a

single AS (autonomous system) More than 40% Gnutella nodes are located within top 10

AS Delayed response

Do we need long-distance neighbors?

Topology Mismatch

Solutions to traffic problem Selective flooding Topology optimization

Avoid cycles Mapping

For each message, how many times it is delivered over a single physical link?

Performance Metrics

Traffic cost Search scope Response time Overhead

Approaches

Location-aware Topology Matching (LTM), INFOCOM 2004

Adaptive Connection Establishment (ACE), ICDCS 2004

LTM

Three main operations1. TTL-2-detector flooding

Message format Short Source IP& timestamp Long Source IP& timestamp, TTL1 IP& timesta

mp d(i,S,v)

Link cost

S N1IP(S),T(S)

IP(S),T(S)N2

IP(N1),T(N1)

d(i,S,1) d(i,S,0)

LTM

Three main operations2. Low productive connection cutting

1. Case1: P receives d(i,S,1) and d(i,S,0)

S

P

N

will-cut list

LTM

Three main operations2. Low productive connection cutting

2. Case2: P receives multiple d(i,S,0)

S

P

N1

N2

LTM

Three main operations2. Low productive connection cutting

3. Case3: P receives one d(i,S,1) and multiple d(i,S,0)

S

P

N1

N2

cut list

LTM

Three main operations3. Source peer probing

S

P

N1

LTM

S

P

N1

N2

Step2.case2 S

P

N1

Step3

LTM

S

P

N1

N2

S

P

N1

N2

Step2.case2Step2.case3

Step2.case1

Step2.case3

LTM

S

P

N1

S

P

N1

Step2.case1Step3

LTM

States

Case1

Step3

Case3

Case2

LTM

Performance Traffic Search scope Overhead

ACE

Step1: Probe link costs with neighbors Build neighbor cost table Exchange neighbors cost table with neighbors

ACE

Step2: Create a minimum spanning tree among each

peer and its neighbors

S

E

F

G

414

15

620

S

E

F

G

414

6

ACE

Step3: Replace neighbors

S

E

F

G

414

6

H

Case1: SH<SG

Case2: GH>SH>SG

Case3: SH>SG,SH>GH

ACE

Depth of optimization (h-neighbor closure)

10 1520

12 148

7

A

B

C

D

E

A->B=10A->D=15

B->E=8D->E=14

E->C=7E->D=14

Total:68

ACE

2-neighbor closure

10 1520

12 148

A

B

C

D

E

A

B

C

D

E7

A->B=10B->E=8E->C=7E->D=14

Total:39

Discussion

Measurement Link cost is not accurate

Link cutting and cycles Heuristic to theoretical support

f (Pn,Tn)=?

Conclusion

Importance Effectiveness vs. cost Future work