DRMT: Disaggregated programmable switching

Sharad Chole, Andy Fingerhut, Sha Ma, Anirudh Sivaraman, Shay Vargaftik, Alon Berger, Gal Mendelson, Mohammad Alizadeh, Shang Tse Chuang, Isaac Keslassy, Ariel Orda, Tom Edsall

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

Abstract

We present dRMT (disaggregated Reconfigurable Match-Action Table), a new architecture for programmable switches. dRMT overcomes two important restrictions of RMT, the predominant pipelinebased architecture for programmable switches: (1) table memory is local to an RMT pipeline stage, implying that memory not used by one stage cannot be reclaimed by another, and (2) RMT is hardwired to always sequentially execute matches followed by actions as packets traverse pipeline stages. We show that these restrictions make it difficult to execute programs efficiently on RMT. dRMT resolves both issues by disaggregating the memory and compute resources of a programmable switch. Specifically, dRMT moves table memories out of pipeline stages and into a centralized pool that is accessible through a crossbar. In addition, dRMT replaces RMT's pipeline stages with a cluster of processors that can execute match and action operations in any order. We show how to schedule a P4 program on dRMT at compile time to guarantee deterministic throughput and latency. We also present a hardware design for dRMT and analyze its feasibility and chip area. Our results show that dRMT can run programs at line rate with fewer processors compared to RMT, and avoids performance cliffs when there are not enough processors to run a program at line rate. dRMT's hardware design incurs a modest increase in chip area relative to RMT, mainly due to the crossbar.

Original languageEnglish (US)
Title of host publicationSIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication
PublisherAssociation for Computing Machinery, Inc
Pages1-14
Number of pages14
ISBN (Electronic)9781450346535
DOIs
StatePublished - Aug 7 2017
Event2017 Conference of the ACM Special Interest Group on Data Communication, SIGCOMM 2017 - Los Angeles, United States
Duration: Aug 21 2017Aug 25 2017

Other

Other2017 Conference of the ACM Special Interest Group on Data Communication, SIGCOMM 2017
CountryUnited States
CityLos Angeles
Period8/21/178/25/17

Fingerprint

Pipelines
Data storage equipment
Switches
hardware
Hardware
guarantee
Throughput
present
resources
performance

Keywords

  • Disagreggation
  • Packet processing
  • Programmable switching
  • RMT

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Signal Processing
  • Electrical and Electronic Engineering
  • Communication

Cite this

Chole, S., Fingerhut, A., Ma, S., Sivaraman, A., Vargaftik, S., Berger, A., ... Edsall, T. (2017). DRMT: Disaggregated programmable switching. In SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication (pp. 1-14). Association for Computing Machinery, Inc. https://doi.org/10.1145/3098822.3098823

DRMT : Disaggregated programmable switching. / Chole, Sharad; Fingerhut, Andy; Ma, Sha; Sivaraman, Anirudh; Vargaftik, Shay; Berger, Alon; Mendelson, Gal; Alizadeh, Mohammad; Chuang, Shang Tse; Keslassy, Isaac; Orda, Ariel; Edsall, Tom.

SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication. Association for Computing Machinery, Inc, 2017. p. 1-14.

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

Chole, S, Fingerhut, A, Ma, S, Sivaraman, A, Vargaftik, S, Berger, A, Mendelson, G, Alizadeh, M, Chuang, ST, Keslassy, I, Orda, A & Edsall, T 2017, DRMT: Disaggregated programmable switching. in SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication. Association for Computing Machinery, Inc, pp. 1-14, 2017 Conference of the ACM Special Interest Group on Data Communication, SIGCOMM 2017, Los Angeles, United States, 8/21/17. https://doi.org/10.1145/3098822.3098823
Chole S, Fingerhut A, Ma S, Sivaraman A, Vargaftik S, Berger A et al. DRMT: Disaggregated programmable switching. In SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication. Association for Computing Machinery, Inc. 2017. p. 1-14 https://doi.org/10.1145/3098822.3098823
Chole, Sharad ; Fingerhut, Andy ; Ma, Sha ; Sivaraman, Anirudh ; Vargaftik, Shay ; Berger, Alon ; Mendelson, Gal ; Alizadeh, Mohammad ; Chuang, Shang Tse ; Keslassy, Isaac ; Orda, Ariel ; Edsall, Tom. / DRMT : Disaggregated programmable switching. SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication. Association for Computing Machinery, Inc, 2017. pp. 1-14
@inproceedings{50ad01b9be9b40eeae4991742f5040eb,
title = "DRMT: Disaggregated programmable switching",
abstract = "We present dRMT (disaggregated Reconfigurable Match-Action Table), a new architecture for programmable switches. dRMT overcomes two important restrictions of RMT, the predominant pipelinebased architecture for programmable switches: (1) table memory is local to an RMT pipeline stage, implying that memory not used by one stage cannot be reclaimed by another, and (2) RMT is hardwired to always sequentially execute matches followed by actions as packets traverse pipeline stages. We show that these restrictions make it difficult to execute programs efficiently on RMT. dRMT resolves both issues by disaggregating the memory and compute resources of a programmable switch. Specifically, dRMT moves table memories out of pipeline stages and into a centralized pool that is accessible through a crossbar. In addition, dRMT replaces RMT's pipeline stages with a cluster of processors that can execute match and action operations in any order. We show how to schedule a P4 program on dRMT at compile time to guarantee deterministic throughput and latency. We also present a hardware design for dRMT and analyze its feasibility and chip area. Our results show that dRMT can run programs at line rate with fewer processors compared to RMT, and avoids performance cliffs when there are not enough processors to run a program at line rate. dRMT's hardware design incurs a modest increase in chip area relative to RMT, mainly due to the crossbar.",
keywords = "Disagreggation, Packet processing, Programmable switching, RMT",
author = "Sharad Chole and Andy Fingerhut and Sha Ma and Anirudh Sivaraman and Shay Vargaftik and Alon Berger and Gal Mendelson and Mohammad Alizadeh and Chuang, {Shang Tse} and Isaac Keslassy and Ariel Orda and Tom Edsall",
year = "2017",
month = "8",
day = "7",
doi = "10.1145/3098822.3098823",
language = "English (US)",
pages = "1--14",
booktitle = "SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - DRMT

T2 - Disaggregated programmable switching

AU - Chole, Sharad

AU - Fingerhut, Andy

AU - Ma, Sha

AU - Sivaraman, Anirudh

AU - Vargaftik, Shay

AU - Berger, Alon

AU - Mendelson, Gal

AU - Alizadeh, Mohammad

AU - Chuang, Shang Tse

AU - Keslassy, Isaac

AU - Orda, Ariel

AU - Edsall, Tom

PY - 2017/8/7

Y1 - 2017/8/7

N2 - We present dRMT (disaggregated Reconfigurable Match-Action Table), a new architecture for programmable switches. dRMT overcomes two important restrictions of RMT, the predominant pipelinebased architecture for programmable switches: (1) table memory is local to an RMT pipeline stage, implying that memory not used by one stage cannot be reclaimed by another, and (2) RMT is hardwired to always sequentially execute matches followed by actions as packets traverse pipeline stages. We show that these restrictions make it difficult to execute programs efficiently on RMT. dRMT resolves both issues by disaggregating the memory and compute resources of a programmable switch. Specifically, dRMT moves table memories out of pipeline stages and into a centralized pool that is accessible through a crossbar. In addition, dRMT replaces RMT's pipeline stages with a cluster of processors that can execute match and action operations in any order. We show how to schedule a P4 program on dRMT at compile time to guarantee deterministic throughput and latency. We also present a hardware design for dRMT and analyze its feasibility and chip area. Our results show that dRMT can run programs at line rate with fewer processors compared to RMT, and avoids performance cliffs when there are not enough processors to run a program at line rate. dRMT's hardware design incurs a modest increase in chip area relative to RMT, mainly due to the crossbar.

AB - We present dRMT (disaggregated Reconfigurable Match-Action Table), a new architecture for programmable switches. dRMT overcomes two important restrictions of RMT, the predominant pipelinebased architecture for programmable switches: (1) table memory is local to an RMT pipeline stage, implying that memory not used by one stage cannot be reclaimed by another, and (2) RMT is hardwired to always sequentially execute matches followed by actions as packets traverse pipeline stages. We show that these restrictions make it difficult to execute programs efficiently on RMT. dRMT resolves both issues by disaggregating the memory and compute resources of a programmable switch. Specifically, dRMT moves table memories out of pipeline stages and into a centralized pool that is accessible through a crossbar. In addition, dRMT replaces RMT's pipeline stages with a cluster of processors that can execute match and action operations in any order. We show how to schedule a P4 program on dRMT at compile time to guarantee deterministic throughput and latency. We also present a hardware design for dRMT and analyze its feasibility and chip area. Our results show that dRMT can run programs at line rate with fewer processors compared to RMT, and avoids performance cliffs when there are not enough processors to run a program at line rate. dRMT's hardware design incurs a modest increase in chip area relative to RMT, mainly due to the crossbar.

KW - Disagreggation

KW - Packet processing

KW - Programmable switching

KW - RMT

UR - http://www.scopus.com/inward/record.url?scp=85029410849&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029410849&partnerID=8YFLogxK

U2 - 10.1145/3098822.3098823

DO - 10.1145/3098822.3098823

M3 - Conference contribution

AN - SCOPUS:85029410849

SP - 1

EP - 14

BT - SIGCOMM 2017 - Proceedings of the 2017 Conference of the ACM Special Interest Group on Data Communication

PB - Association for Computing Machinery, Inc

ER -