An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs

Yang Xu, Zhaobo Liu, Zhuoyuan Zhang, H. Jonathan Chao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The emergence of new network applications like network intrusion detection system, packet-level accounting, and load-balancing requires packet classification to report all matched rules, instead of only the best matched rule. Although several schemes have been proposed recently to address the multi-match packet classification problem, most of them require either huge memory or expensive Ternary Content Addressable Memory (TCAM) to store the intermediate data structure, or suffer from steep performance degradation under certain types of classifiers. In this paper, we decompose the operation of multi-match packet classification from the complicated multi-dimensional search to several single-dimensional searches, and present an asynchronous pipeline architecture based on a signature tree structure to combine the intermediate results returned from single-dimensional searches. By spreading edges of the signature tree in multiple hash tables at different stages of the pipeline, the pipeline can achieve a high throughput via the inter-stage parallel access to hash tables. To exploit further intra-stage parallelism, two edge-grouping algorithms are designed to evenly divide the edges associated with each stage into multiple work-conserving hash tables with minimum overhead. Extensive simulation using realistic classifiers and traffic traces shows that the proposed pipeline architecture outperforms HyperCut and B2PC schemes in classification speed by at least one order of magnitude, while with a similar storage requirement. Particularly, with different types of classifiers of 4K rules, the proposed pipeline architecture is able to achieve a throughput between 19.5 Gbps and 91 Gbps.

Original languageEnglish (US)
Title of host publicationANCS'09: Symposium on Architecture for Networking and Communications Systems
Pages189-198
Number of pages10
DOIs
StatePublished - 2009
Event2009 Symposium on Architecture for Networking and Communications Systems, ANCS'09 - Princeton, NJ, United States
Duration: Oct 19 2009Oct 20 2009

Other

Other2009 Symposium on Architecture for Networking and Communications Systems, ANCS'09
CountryUnited States
CityPrinceton, NJ
Period10/19/0910/20/09

Fingerprint

Pipelines
Throughput
Data storage equipment
Classifiers
Associative storage
Intrusion detection
Resource allocation
Data structures
Degradation

Keywords

  • hash table
  • packet classification
  • signature tree
  • TCAM

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Hardware and Architecture
  • Electrical and Electronic Engineering

Cite this

Xu, Y., Liu, Z., Zhang, Z., & Chao, H. J. (2009). An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs. In ANCS'09: Symposium on Architecture for Networking and Communications Systems (pp. 189-198) https://doi.org/10.1145/1882486.1882537

An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs. / Xu, Yang; Liu, Zhaobo; Zhang, Zhuoyuan; Chao, H. Jonathan.

ANCS'09: Symposium on Architecture for Networking and Communications Systems. 2009. p. 189-198.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Xu, Y, Liu, Z, Zhang, Z & Chao, HJ 2009, An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs. in ANCS'09: Symposium on Architecture for Networking and Communications Systems. pp. 189-198, 2009 Symposium on Architecture for Networking and Communications Systems, ANCS'09, Princeton, NJ, United States, 10/19/09. https://doi.org/10.1145/1882486.1882537
Xu Y, Liu Z, Zhang Z, Chao HJ. An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs. In ANCS'09: Symposium on Architecture for Networking and Communications Systems. 2009. p. 189-198 https://doi.org/10.1145/1882486.1882537
Xu, Yang ; Liu, Zhaobo ; Zhang, Zhuoyuan ; Chao, H. Jonathan. / An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs. ANCS'09: Symposium on Architecture for Networking and Communications Systems. 2009. pp. 189-198
@inproceedings{44201c8425c6484082fcf14d9089977b,
title = "An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs",
abstract = "The emergence of new network applications like network intrusion detection system, packet-level accounting, and load-balancing requires packet classification to report all matched rules, instead of only the best matched rule. Although several schemes have been proposed recently to address the multi-match packet classification problem, most of them require either huge memory or expensive Ternary Content Addressable Memory (TCAM) to store the intermediate data structure, or suffer from steep performance degradation under certain types of classifiers. In this paper, we decompose the operation of multi-match packet classification from the complicated multi-dimensional search to several single-dimensional searches, and present an asynchronous pipeline architecture based on a signature tree structure to combine the intermediate results returned from single-dimensional searches. By spreading edges of the signature tree in multiple hash tables at different stages of the pipeline, the pipeline can achieve a high throughput via the inter-stage parallel access to hash tables. To exploit further intra-stage parallelism, two edge-grouping algorithms are designed to evenly divide the edges associated with each stage into multiple work-conserving hash tables with minimum overhead. Extensive simulation using realistic classifiers and traffic traces shows that the proposed pipeline architecture outperforms HyperCut and B2PC schemes in classification speed by at least one order of magnitude, while with a similar storage requirement. Particularly, with different types of classifiers of 4K rules, the proposed pipeline architecture is able to achieve a throughput between 19.5 Gbps and 91 Gbps.",
keywords = "hash table, packet classification, signature tree, TCAM",
author = "Yang Xu and Zhaobo Liu and Zhuoyuan Zhang and Chao, {H. Jonathan}",
year = "2009",
doi = "10.1145/1882486.1882537",
language = "English (US)",
isbn = "9781605586304",
pages = "189--198",
booktitle = "ANCS'09: Symposium on Architecture for Networking and Communications Systems",

}

TY - GEN

T1 - An ultra high throughput and memory efficient pipeline architecture for multi-match packet classification without TCAMs

AU - Xu, Yang

AU - Liu, Zhaobo

AU - Zhang, Zhuoyuan

AU - Chao, H. Jonathan

PY - 2009

Y1 - 2009

N2 - The emergence of new network applications like network intrusion detection system, packet-level accounting, and load-balancing requires packet classification to report all matched rules, instead of only the best matched rule. Although several schemes have been proposed recently to address the multi-match packet classification problem, most of them require either huge memory or expensive Ternary Content Addressable Memory (TCAM) to store the intermediate data structure, or suffer from steep performance degradation under certain types of classifiers. In this paper, we decompose the operation of multi-match packet classification from the complicated multi-dimensional search to several single-dimensional searches, and present an asynchronous pipeline architecture based on a signature tree structure to combine the intermediate results returned from single-dimensional searches. By spreading edges of the signature tree in multiple hash tables at different stages of the pipeline, the pipeline can achieve a high throughput via the inter-stage parallel access to hash tables. To exploit further intra-stage parallelism, two edge-grouping algorithms are designed to evenly divide the edges associated with each stage into multiple work-conserving hash tables with minimum overhead. Extensive simulation using realistic classifiers and traffic traces shows that the proposed pipeline architecture outperforms HyperCut and B2PC schemes in classification speed by at least one order of magnitude, while with a similar storage requirement. Particularly, with different types of classifiers of 4K rules, the proposed pipeline architecture is able to achieve a throughput between 19.5 Gbps and 91 Gbps.

AB - The emergence of new network applications like network intrusion detection system, packet-level accounting, and load-balancing requires packet classification to report all matched rules, instead of only the best matched rule. Although several schemes have been proposed recently to address the multi-match packet classification problem, most of them require either huge memory or expensive Ternary Content Addressable Memory (TCAM) to store the intermediate data structure, or suffer from steep performance degradation under certain types of classifiers. In this paper, we decompose the operation of multi-match packet classification from the complicated multi-dimensional search to several single-dimensional searches, and present an asynchronous pipeline architecture based on a signature tree structure to combine the intermediate results returned from single-dimensional searches. By spreading edges of the signature tree in multiple hash tables at different stages of the pipeline, the pipeline can achieve a high throughput via the inter-stage parallel access to hash tables. To exploit further intra-stage parallelism, two edge-grouping algorithms are designed to evenly divide the edges associated with each stage into multiple work-conserving hash tables with minimum overhead. Extensive simulation using realistic classifiers and traffic traces shows that the proposed pipeline architecture outperforms HyperCut and B2PC schemes in classification speed by at least one order of magnitude, while with a similar storage requirement. Particularly, with different types of classifiers of 4K rules, the proposed pipeline architecture is able to achieve a throughput between 19.5 Gbps and 91 Gbps.

KW - hash table

KW - packet classification

KW - signature tree

KW - TCAM

UR - http://www.scopus.com/inward/record.url?scp=78650959885&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650959885&partnerID=8YFLogxK

U2 - 10.1145/1882486.1882537

DO - 10.1145/1882486.1882537

M3 - Conference contribution

SN - 9781605586304

SP - 189

EP - 198

BT - ANCS'09: Symposium on Architecture for Networking and Communications Systems

ER -