View in EDS HTML Full Text PDF Full Text

Towards connection-scalable RNIC architecture.

Saved in:
Bibliographic Details
Title: Towards connection-scalable RNIC architecture.
Authors: Kang, Ning1,2 (AUTHOR), Wang, Zhan1 (AUTHOR) wangzhan@ncic.ac.cn, Yang, Fan1 (AUTHOR), Ma, Xiaoxiao1,2 (AUTHOR), Ma, Zhenlong1,2 (AUTHOR), Yuan, Guojun1 (AUTHOR), Tan, Guangming1 (AUTHOR)
Source: Journal of Supercomputing. Jul2024, Vol. 80 Issue 10, p14548-14572. 25p.
Subjects: TCP/IP, Kernel operating systems, Footprints, Architectural design, Scalability
Abstract: Remote Direct Memory Access (RDMA) is a widely adopted optimization strategy in datacenter networking that surpasses traditional kernel-based TCP/IP networking through mechanisms such as kernel bypass and hardware offloading. However, RDMA also faces a scalability challenge with regard to connection management due to limited on-chip memory capacity in the RDMA Network Interface Card (RNIC). This necessitates the storage of connection context within RNIC's memory and induces considerable performance degradation when maintaining a large number of connections. In this paper, we propose a novel RNIC microarchitecture design that achieves peak performance and scales well with the number of connections. First, we model RNIC and identify two key factors that degrade performance when the number of connections grows large: head-of-line blocking when accessing the connection context and connection context dependency in transmission processing. To address the head-of-line blocking problem, we then combine a non-blocking connection requester and connection context management module to process prepared connections first, which achieves peak message rate when the number of connections grows large. Besides, to eliminate connection context dependency in RNIC, we deploy a latency-hiding connection context scheduling strategy, maintaining low latency when the number of connections increases. We implement and evaluate our design, demonstrating its successful maintenance of peak message rate (66.4 Mop/s) and low latency (3.89 µs) while scaling to over 50,000 connections with less on-chip memory footprint. [ABSTRACT FROM AUTHOR]
Copyright of Journal of Supercomputing is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
Text:
  Availability: 1
Header DbId: egs
DbLabel: Engineering Source
An: 177776499
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Towards connection-scalable RNIC architecture.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AR" term="%22Kang%2C+Ning%22">Kang, Ning</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Wang%2C+Zhan%22">Wang, Zhan</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> wangzhan@ncic.ac.cn</i><br /><searchLink fieldCode="AR" term="%22Yang%2C+Fan%22">Yang, Fan</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Ma%2C+Xiaoxiao%22">Ma, Xiaoxiao</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Ma%2C+Zhenlong%22">Ma, Zhenlong</searchLink><relatesTo>1,2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Yuan%2C+Guojun%22">Yuan, Guojun</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Tan%2C+Guangming%22">Tan, Guangming</searchLink><relatesTo>1</relatesTo> (AUTHOR)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22Journal+of+Supercomputing%22">Journal of Supercomputing</searchLink>. Jul2024, Vol. 80 Issue 10, p14548-14572. 25p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: <searchLink fieldCode="DE" term="%22TCP%2FIP%22">TCP/IP</searchLink><br /><searchLink fieldCode="DE" term="%22Kernel+operating+systems%22">Kernel operating systems</searchLink><br /><searchLink fieldCode="DE" term="%22Footprints%22">Footprints</searchLink><br /><searchLink fieldCode="DE" term="%22Architectural+design%22">Architectural design</searchLink><br /><searchLink fieldCode="DE" term="%22Scalability%22">Scalability</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Remote Direct Memory Access (RDMA) is a widely adopted optimization strategy in datacenter networking that surpasses traditional kernel-based TCP/IP networking through mechanisms such as kernel bypass and hardware offloading. However, RDMA also faces a scalability challenge with regard to connection management due to limited on-chip memory capacity in the RDMA Network Interface Card (RNIC). This necessitates the storage of connection context within RNIC's memory and induces considerable performance degradation when maintaining a large number of connections. In this paper, we propose a novel RNIC microarchitecture design that achieves peak performance and scales well with the number of connections. First, we model RNIC and identify two key factors that degrade performance when the number of connections grows large: head-of-line blocking when accessing the connection context and connection context dependency in transmission processing. To address the head-of-line blocking problem, we then combine a non-blocking connection requester and connection context management module to process prepared connections first, which achieves peak message rate when the number of connections grows large. Besides, to eliminate connection context dependency in RNIC, we deploy a latency-hiding connection context scheduling strategy, maintaining low latency when the number of connections increases. We implement and evaluate our design, demonstrating its successful maintenance of peak message rate (66.4 Mop/s) and low latency (3.89 µs) while scaling to over 50,000 connections with less on-chip memory footprint. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: <i>Copyright of Journal of Supercomputing is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=177776499
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1007/s11227-024-05991-4
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 25
        StartPage: 14548
    Subjects:
      – SubjectFull: TCP/IP
        Type: general
      – SubjectFull: Kernel operating systems
        Type: general
      – SubjectFull: Footprints
        Type: general
      – SubjectFull: Architectural design
        Type: general
      – SubjectFull: Scalability
        Type: general
    Titles:
      – TitleFull: Towards connection-scalable RNIC architecture.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Kang, Ning
      – PersonEntity:
          Name:
            NameFull: Wang, Zhan
      – PersonEntity:
          Name:
            NameFull: Yang, Fan
      – PersonEntity:
          Name:
            NameFull: Ma, Xiaoxiao
      – PersonEntity:
          Name:
            NameFull: Ma, Zhenlong
      – PersonEntity:
          Name:
            NameFull: Yuan, Guojun
      – PersonEntity:
          Name:
            NameFull: Tan, Guangming
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 07
              Text: Jul2024
              Type: published
              Y: 2024
          Identifiers:
            – Type: issn-print
              Value: 09208542
          Numbering:
            – Type: volume
              Value: 80
            – Type: issue
              Value: 10
          Titles:
            – TitleFull: Journal of Supercomputing
              Type: main
ResultId 1