Measuring and evaluating TCP splitting for cloud services

Abhinav Pathak, Y. Angela Wang, Cheng Huang, Albert Greenberg, Y. Charlie Hu, Randy Kern, Jin Li, Keith Ross

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

    Abstract

    In this paper, we examine the benefits of split-TCP proxies, deployed in an operational world-wide network, for accelerating cloud services. We consider a fraction of a network consisting of a large number of satellite datacenters, which host split-TCP proxies, and a smaller number of mega datacenters, which ultimately perform computation or provide storage. Using web search as an exemplary case study, our detailed measurements reveal that a vanilla TCP splitting solution deployed at the satellite DCs reduces the 95th percentile of latency by as much as 43% when compared to serving queries directly from the mega DCs. Through careful dissection of the measurement results, we characterize how individual components, including proxy stacks, network protocols, packet losses and network load, can impact the latency. Finally, we shed light on further optimizations that can fully realize the potential of the TCP splitting solution.

    Original languageEnglish (US)
    Title of host publicationPassive and Active Measurement - 11th International Conference, PAM 2010, Proceedings
    PublisherSpringer Verlag
    Pages41-50
    Number of pages10
    Volume6032 LNCS
    ISBN (Print)9783642123337
    StatePublished - 2010
    Event11th International Conference on Passive and Active Network Measurement, PAM 2010 - Zurich, Switzerland
    Duration: Apr 7 2010Apr 9 2010

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume6032 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Other

    Other11th International Conference on Passive and Active Network Measurement, PAM 2010
    CountrySwitzerland
    CityZurich
    Period4/7/104/9/10

    Fingerprint

    Satellites
    Dissection
    Latency
    Packet networks
    Packet loss
    Network Protocols
    Percentile
    Web Search
    Packet Loss
    Network protocols
    Query
    Optimization

    ASJC Scopus subject areas

    • Theoretical Computer Science
    • Computer Science(all)

    Cite this

    Pathak, A., Wang, Y. A., Huang, C., Greenberg, A., Hu, Y. C., Kern, R., ... Ross, K. (2010). Measuring and evaluating TCP splitting for cloud services. In Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings (Vol. 6032 LNCS, pp. 41-50). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6032 LNCS). Springer Verlag.

    Measuring and evaluating TCP splitting for cloud services. / Pathak, Abhinav; Wang, Y. Angela; Huang, Cheng; Greenberg, Albert; Hu, Y. Charlie; Kern, Randy; Li, Jin; Ross, Keith.

    Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings. Vol. 6032 LNCS Springer Verlag, 2010. p. 41-50 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6032 LNCS).

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

    Pathak, A, Wang, YA, Huang, C, Greenberg, A, Hu, YC, Kern, R, Li, J & Ross, K 2010, Measuring and evaluating TCP splitting for cloud services. in Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings. vol. 6032 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6032 LNCS, Springer Verlag, pp. 41-50, 11th International Conference on Passive and Active Network Measurement, PAM 2010, Zurich, Switzerland, 4/7/10.
    Pathak A, Wang YA, Huang C, Greenberg A, Hu YC, Kern R et al. Measuring and evaluating TCP splitting for cloud services. In Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings. Vol. 6032 LNCS. Springer Verlag. 2010. p. 41-50. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    Pathak, Abhinav ; Wang, Y. Angela ; Huang, Cheng ; Greenberg, Albert ; Hu, Y. Charlie ; Kern, Randy ; Li, Jin ; Ross, Keith. / Measuring and evaluating TCP splitting for cloud services. Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings. Vol. 6032 LNCS Springer Verlag, 2010. pp. 41-50 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{72bd0b53d1e641fb9b0a5e2a07e06627,
    title = "Measuring and evaluating TCP splitting for cloud services",
    abstract = "In this paper, we examine the benefits of split-TCP proxies, deployed in an operational world-wide network, for accelerating cloud services. We consider a fraction of a network consisting of a large number of satellite datacenters, which host split-TCP proxies, and a smaller number of mega datacenters, which ultimately perform computation or provide storage. Using web search as an exemplary case study, our detailed measurements reveal that a vanilla TCP splitting solution deployed at the satellite DCs reduces the 95th percentile of latency by as much as 43{\%} when compared to serving queries directly from the mega DCs. Through careful dissection of the measurement results, we characterize how individual components, including proxy stacks, network protocols, packet losses and network load, can impact the latency. Finally, we shed light on further optimizations that can fully realize the potential of the TCP splitting solution.",
    author = "Abhinav Pathak and Wang, {Y. Angela} and Cheng Huang and Albert Greenberg and Hu, {Y. Charlie} and Randy Kern and Jin Li and Keith Ross",
    year = "2010",
    language = "English (US)",
    isbn = "9783642123337",
    volume = "6032 LNCS",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer Verlag",
    pages = "41--50",
    booktitle = "Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings",
    address = "Germany",

    }

    TY - GEN

    T1 - Measuring and evaluating TCP splitting for cloud services

    AU - Pathak, Abhinav

    AU - Wang, Y. Angela

    AU - Huang, Cheng

    AU - Greenberg, Albert

    AU - Hu, Y. Charlie

    AU - Kern, Randy

    AU - Li, Jin

    AU - Ross, Keith

    PY - 2010

    Y1 - 2010

    N2 - In this paper, we examine the benefits of split-TCP proxies, deployed in an operational world-wide network, for accelerating cloud services. We consider a fraction of a network consisting of a large number of satellite datacenters, which host split-TCP proxies, and a smaller number of mega datacenters, which ultimately perform computation or provide storage. Using web search as an exemplary case study, our detailed measurements reveal that a vanilla TCP splitting solution deployed at the satellite DCs reduces the 95th percentile of latency by as much as 43% when compared to serving queries directly from the mega DCs. Through careful dissection of the measurement results, we characterize how individual components, including proxy stacks, network protocols, packet losses and network load, can impact the latency. Finally, we shed light on further optimizations that can fully realize the potential of the TCP splitting solution.

    AB - In this paper, we examine the benefits of split-TCP proxies, deployed in an operational world-wide network, for accelerating cloud services. We consider a fraction of a network consisting of a large number of satellite datacenters, which host split-TCP proxies, and a smaller number of mega datacenters, which ultimately perform computation or provide storage. Using web search as an exemplary case study, our detailed measurements reveal that a vanilla TCP splitting solution deployed at the satellite DCs reduces the 95th percentile of latency by as much as 43% when compared to serving queries directly from the mega DCs. Through careful dissection of the measurement results, we characterize how individual components, including proxy stacks, network protocols, packet losses and network load, can impact the latency. Finally, we shed light on further optimizations that can fully realize the potential of the TCP splitting solution.

    UR - http://www.scopus.com/inward/record.url?scp=84963791095&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84963791095&partnerID=8YFLogxK

    M3 - Conference contribution

    SN - 9783642123337

    VL - 6032 LNCS

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 41

    EP - 50

    BT - Passive and Active Measurement - 11th International Conference, PAM 2010, Proceedings

    PB - Springer Verlag

    ER -