39
Packet Classification on Multiple Fields 참참 참참 : Pankaj Gupta and Nick McKeown SigComm 1999

Packet Classification on Multiple Fields

  • Upload
    iago

  • View
    85

  • Download
    1

Embed Size (px)

DESCRIPTION

Packet Classification on Multiple Fields. 참고 논문: Pankaj Gupta and Nick McKeown SigComm 1999. IP Lookup. Longest-prefix address lookup. Rule: Each Prefix. Action: Next hop. Classifier: Forwarding table. Outline. Packet Classification Application, Information, Characteristics Design Goal - PowerPoint PPT Presentation

Citation preview

Page 1: Packet Classification on Multiple Fields

Packet Classification on Multiple Fields

참고 논문 : Pankaj Gupta and Nick McKeownSigComm 1999

Page 2: Packet Classification on Multiple Fields

2

IP Lookup

Longest-prefix address lookup

Rule: Each Prefix

Action: Next hop Classifier: Forwarding table

Page 3: Packet Classification on Multiple Fields

3

Outline

Packet Classification Application, Information,

Characteristics

Design GoalPrevious WorkRecursive Flow Classification (RFC)

Page 4: Packet Classification on Multiple Fields

4

Packet classification(Application)

Packet Filtering Deny all traffic from ISP3

destined to E2

Accounting & Billing Treat all video traffic to

E1 as highest priority and perform accounting for the traffic sent this way

Page 5: Packet Classification on Multiple Fields

5

Packet classification(Application)

Page 6: Packet Classification on Multiple Fields

6

Packet Header

http://vulcan.ee.iastate.edu/~dougj/class/580/index.htm

Page 7: Packet Classification on Multiple Fields

7

Example

Page 8: Packet Classification on Multiple Fields

8

Example (Cont.)

Page 9: Packet Classification on Multiple Fields

9

Definition of Packet Classification

Each rule Specifies a class Based on criterion on F fields Associates with an identifier, classID i-th component of the rule R, R[i]

A regular expression on the i-th filed of the packet header

Page 10: Packet Classification on Multiple Fields

10

Characteristics of P-C

793 packet classifiers 101 different ISP and enterprise networksTotal 41505 rules? Data representative? Conclusion: Trivial or non-trivial?

Page 11: Packet Classification on Multiple Fields

11

Distribution of total number of rules per classifier

Page 12: Packet Classification on Multiple Fields

12

Characteristics 1

The classifiers do not contain a large number of rules. 0.7% of the classifiers contain more

than 1000 rules Mean number: 50 rules

Page 13: Packet Classification on Multiple Fields

13

Characteristics 2

The syntax allows a maximum of 8 fields to be specified: Source/destination Network-Layer

address(32-bits) Source/destination Transport-layer

port number(16-bits for TCP/UDP) Type-of-service field(8-bits) Protocol field (8-bits) Transport-Layer protocol flags(8-bits)

Page 14: Packet Classification on Multiple Fields

14

Characteristics 3-4

Transport-Layer protocol field is restricted to a small set of values: TCP, UDP, ICMP, IGMP, (E)IGRP, GRE

and IPINIP or wildcard

Transport-layer fields Many (10.2%) are range

specifications

Page 15: Packet Classification on Multiple Fields

15

Characteristics 5-6

14% of all the classifiers had a rule with a non-contiguous mask.Many different rules share a number of field specifications.

Page 16: Packet Classification on Multiple Fields

16

Characteristics 7

Redundant Backward redundancy

Rule T appears earlier than Rule R, and R is a subset of T.

Forward redundancy Rule T apperas after R

R is a subset of T R, T have the same action Rules inbetween R and T

The same action Disjoint from R.

8% of the rules were redundant.

Page 17: Packet Classification on Multiple Fields

17

Goals

Fast enough Matching on arbitrary fieldsSupport general classification rules Prefixes, operators(like range, less than,

greater than…) and wildcards.

Suitable for software and hardware implementationMemory efficientScalabilityFor steady classifier

Page 18: Packet Classification on Multiple Fields

18

Previous Work

Sequential evaluationsGrid of TriesCrossproductingBit-level parallelismTCAM

Page 19: Packet Classification on Multiple Fields

19

Abstract

The point location problem in multidimensional spaceFind the enclosing region of a point, given a set of regions.Complexity O(logn) in time with O(nF) space O(logF-1n) time with O(n) space

Page 20: Packet Classification on Multiple Fields

20

Structure of classifiers

Page 21: Packet Classification on Multiple Fields

21

Overlapping is small

For the biggest classifier with 1734 rules, the number of distinct overlapping regions in four dimensions to be 4316, compared to a worst possible case of approximately 1013.

Page 22: Packet Classification on Multiple Fields

22

Recursive Flow Classification

Page 23: Packet Classification on Multiple Fields

23

Chunks of Packet header

Page 24: Packet Classification on Multiple Fields

24

Packet flow in RFC

Page 25: Packet Classification on Multiple Fields

25

Phase number =3

Page 26: Packet Classification on Multiple Fields

26

Phase number =4

Page 27: Packet Classification on Multiple Fields

27

Select Phase number

Combine those chunks together which have the most “correlation”.Combine as many chunks as possible without causing unreasonable memory consumption.Best case: P=3 Tree B P=4 Tree A

Page 28: Packet Classification on Multiple Fields

28

Performance (Storage requirement for

P=2)

Page 29: Packet Classification on Multiple Fields

29

Performance (Storage requirement for P=3)

Page 30: Packet Classification on Multiple Fields

30

Performance (Storage requirement for P=4)

Page 31: Packet Classification on Multiple Fields

31

Preprocess Time

Page 32: Packet Classification on Multiple Fields

32

Hardware implementation

Page 33: Packet Classification on Multiple Fields

33

Larger classifiers IConcatenating the classifiers belonging to the

same network.

Page 34: Packet Classification on Multiple Fields

34

Larger classifiers II

Concatenate all the classifiers of a few (up to ten) different networks. RFC frequently runs into storage

problems for classifiers with more than 6000 rules.

Page 35: Packet Classification on Multiple Fields

35

Variations

Process a larger number of fields in each packet header.Use available fast lookup algorithmsUse Adjacency group

Page 36: Packet Classification on Multiple Fields

36

Adjacency Groups

Two rules (R, S)are considered adjacent R appears first The same Action All but one field have the same

specification All rules in between R and S

Either have the same action Disjoint from R

Page 37: Packet Classification on Multiple Fields

37

An example

Page 38: Packet Classification on Multiple Fields

38

Storage (Adjacency groups)

Page 39: Packet Classification on Multiple Fields

39

Comments

Trade off? Memory, Speed, Dynamic Change, etc

Application and Demand orientedFurther discussion? [email protected]

Thank you!