Top Banner
SCALABL E Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone R U O I T N G FLAT NAMES ON
37

SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

Dec 15, 2015

Download

Documents

Josue Elmer
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

SCALABLE

Ankit Singla, P. Brighten GodfreyKevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone

R UO ITN

G

FLAT NAMESFLAT NAMESON

Page 2: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

Scalable Routing

NUMBER OF NODES = n

NOT VERY SCALABLE!

STATE

=

SHORTEST-PATH ROUTING

?ROU

TIN

G

STAT

E/N

OD

E

Page 3: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

3

Scalable Routing

STRUCTURED

TOPOLOGIES

4, 11

6, 20

But, what about general topologies?

Page 4: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

4

Scalable Routing

HIERARCHY

• Inefficient• Location-dependent

addressings

t

Page 5: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

5

Why Not DNS/DHTs?

Arbitrarily long paths!

WHERE IS A?

A

RIGHT NEXT TO YOU, SILLY!

Page 6: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

6

Local Name Resolution

• Why local name-resolution matters– Google: 400ms delay = 0.59% searches/user– Time for name-resolution is a significant fraction

of latency for short flows (which are the majority)– Fate sharing property

Page 7: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

7

• Arbitrary names – independent of location• Useful primitive– Mobility– Security– Simplified management

Flat Names

Page 8: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

8

Another Way to Scale• Compact Routing– [Thorup & Zwick ’01, Abraham et al. ’04]– Idea: accept (slightly) longer paths, get lower state

– State => Stretch

(stretch = )

Page 9: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

state; stretch = 3

state; stretch = k

.

.

.

state; stretch = 5

Compact Routing

NUMBER OF NODES = n

state; stretch = 1

ROU

TIN

G

STAT

E/N

OD

E

9

Page 10: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

Algorithm != Protocol

NOT DISTRIBUTEDNOT DYNAMIC

10

Page 11: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

11

Disco

Disco: Distributed Compact Routing

• Scalable: state* per node• Efficient: stretch • Flat-names: Route over arbitrary names

*Assumption: partial source routes are of size O(polylog(n))

3..

Page 12: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

12

Disco

Local Name Resolution

Name Resolution

NDDisco: Name-Dependent Distributed Compact RoutingNDDisco: Name-Dependent

Distributed Compact Routing

Page 13: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

13

NDDisco

sAddr (t) = (L(t), b, t)

t

L(t)

) random landmarks (enable approximately shortest paths)

b

Page 14: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

14

NDDisco

s

t'

L(t')

For nearby nodes, the approximation can be poor!

Page 15: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

15

NDDisco

s

t'

L(t)

) random landmarks ) node vicinity

t

Page 16: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

16

NDDisco

s

L(t)

Destination can help!

t

Page 17: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

17

Disco

Local Name Resolution

Name Resolution

NDDisco: Name-Dependent Distributed Compact Routing

Name Resolution

Page 18: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

18

Name Resolution• Name -> Address

mappings stored in DHT over landmarks

• Similar to prior work [BVR, S4]

Arbitrarily high stretch for address lookup!

st

Page 19: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

19

Disco

Local Name Resolution

Name Resolution

NDDisco: Name-Dependent Distributed Compact Routing

Local Name Resolution

Page 20: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

20

Local Name Resolution• Scatter t’s address to )

random nodes

• Then each node s has t’s address somewhere in its vicinity w.h.p.

• ‘Scattering’ addresses?• Finding addresses?

ts

?L(t)

Smells like hash functions!

Page 21: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

21

Sloppy Groups

• G(t) = nodes w for which first k bits of h(w) match h(t)• Set k such that |G(t)| = () • Groups are sloppy because nodes’ estimates of ‘n’ vary!

Increasing h()

EVERYONE AGREES ON THIS

PARTt

G(t)

Page 22: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

22

Address Dissemination

• Requirements:– Low diameter for quick dissemination of addresses– Low messaging overhead– Fault tolerance

Increasing h()

Page 23: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

23

Address Dissemination

• Predecessor, successor and small O(1) number of fingers• Distance-vector style dissemination of addresses

Increasing h()

Page 24: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

24

1: Overlay Topology

• Predecessor, successor and small O(1) number of fingers– Fault tolerance: Use landmarks to fix overlay when broken– Miserly messaging: each node hears each address O(1) times– We don’t care about routing, only diameter, which is small!

Increasing h()

Page 25: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

25

2: Dissemination Protocol

• Disseminate new addresses learnt to overlay neighbors• Record where each address was learnt from (distance vector-like)• Always forward away from the direction received from

– Avoids count-to-infinity problem because of ordered space

Increasing h()

Page 26: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

26

Handling Failures: Example

No soft-state required!Update only on changes

Increasing h()

GET NEW NEIGHBOR FROM LANDMARK

WITHDRAW ADDRESSESLEARNT FROM X X

Page 27: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

27

Routing in Disco

w(t)

• s computes h(t)• w(t) = Longest-prefix

match for h(t) in Vicinity(s)

First : s w(t) L(t) tLater: s L(t) t

st

L(t)

Page 28: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

28

Protocol Messages• Estimating ‘n’: synopses

exchanged with neighbors• Landmark and Vicinity

updates: path vector style• Sloppy-group bootstrap

through landmarks• Address-dissemination:

distance-vector style

s

Page 29: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

29

• State at a node– Vicinity: ) closest nodes– Landmarks: ) random landmarks– Address-Groups: ) overlay neighbors

•Stretch– For first packet <= 7– For later packets <= 3

Summary

Page 30: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

30

Evaluation

• Protocols– Disco– NDDisco– S4 [Mao et al ‘07]– VRR [Caesar et al ‘06]

• refer to paper

• Topologies– Router-level Internet

topology (n = 192,244)– AS-level Internet topology

(n = 30,610)– Geometric random

graphs, mean degree 8– G(n, m), mean degree 8

Page 31: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

StateSHORTEST PATH

PROTOCOL

31

NDDISCO

S4

GEOMETRIC RANDOM GRAPHS OF INCREASING SIZE

Page 32: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

32

State

LONG TAIL

S4

DISCO

NDDISCO

ROUTER-LEVEL INTERNET TOPOLOGY

Page 33: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

33

First-Packet Stretch

GEOMETRIC RANDOM GRAPH16384-NODES

ROUTER-LEVEL INTERNETTOPOLOGY

72

LARGE FIRST-PACKETSTRETCH

DISCO S4

S4

DISCO

LIMITED BY SMALL PATH-LENGTHS IN

TOPOLOGY

Page 34: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

34

Later-Packets Stretch

GEOMETRIC RANDOM GRAPH16384-NODES

ROUTER-LEVEL INTERNETTOPOLOGY

DISCO < 1.2NEAR-IDENTICAL

S4DISCO

Page 35: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

Are the Landmarks Clobbered?

AS-LEVEL INTERNET TOPOLOGY

35

~4 IN EVERY 10,000 EDGES SEE MORE

CONGESTION

DISCO

Page 36: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

36

Conclusion

• Disco routes over flat names with O(1) stretch and state

• What next?– Policy support?– Applications

Page 37: SCALABLESCALABLE Ankit Singla, P. Brighten Godfrey Kevin Fall, Sylvia Ratnasamy, Gianluca Iannaccone ON.

Thank You

Questions?