Stochastic fluid models for cache clusters

Florence Clévenot, Philippe Nain, Keith Ross

    Research output: Contribution to journalArticle

    Abstract

    Clusters of Web caches are extensively used by different types of organizations, including companies, universities, ISPs, and CDNs. To model Web caches, we must account for two types of stochastic events: objects being pulled into/out of the cache cluster at random times, and caches going up and down at random times. Detailed stochastic models of such complex systems quickly become intractable. In this paper we propose a stochastic fluid model which captures the salient characteristics of a cache cluster. The stochastic fluid model replaces the object arrivals to the cluster and departures (object modification/expiration) with a fluid flow, but maintains the up/down dynamics of the original system. The model can be applied to a variety of cluster routing policies, and provides a simple means to estimate the hit rate. We compare the results of the stochastic fluid model with that of a simulation of the real system. We find the fluid model to not only be a close approximation, but also to exhibit the key qualitative properties of the original system. We conclude that stochastic fluid models show great potential in modeling a variety of content distribution systems.

    Original languageEnglish (US)
    Pages (from-to)1-18
    Number of pages18
    JournalPerformance Evaluation
    Volume59
    Issue number1
    DOIs
    StatePublished - Jan 2005

    Fingerprint

    Fluid Model
    Cache
    Stochastic Model
    Fluids
    Content Distribution
    Qualitative Properties
    Distribution System
    Hits
    Fluid Flow
    Complex Systems
    Routing
    Stochastic models
    Large scale systems
    Flow of fluids
    Approximation
    Modeling
    Model
    Estimate
    Object
    Simulation

    Keywords

    • Fluid models
    • Palm calculus
    • Stochastic processes
    • Web caching

    ASJC Scopus subject areas

    • Computer Networks and Communications
    • Software
    • Modeling and Simulation
    • Statistics and Probability

    Cite this

    Stochastic fluid models for cache clusters. / Clévenot, Florence; Nain, Philippe; Ross, Keith.

    In: Performance Evaluation, Vol. 59, No. 1, 01.2005, p. 1-18.

    Research output: Contribution to journalArticle

    Clévenot, Florence ; Nain, Philippe ; Ross, Keith. / Stochastic fluid models for cache clusters. In: Performance Evaluation. 2005 ; Vol. 59, No. 1. pp. 1-18.
    @article{5610f829eb92489aa1cb3691312e766d,
    title = "Stochastic fluid models for cache clusters",
    abstract = "Clusters of Web caches are extensively used by different types of organizations, including companies, universities, ISPs, and CDNs. To model Web caches, we must account for two types of stochastic events: objects being pulled into/out of the cache cluster at random times, and caches going up and down at random times. Detailed stochastic models of such complex systems quickly become intractable. In this paper we propose a stochastic fluid model which captures the salient characteristics of a cache cluster. The stochastic fluid model replaces the object arrivals to the cluster and departures (object modification/expiration) with a fluid flow, but maintains the up/down dynamics of the original system. The model can be applied to a variety of cluster routing policies, and provides a simple means to estimate the hit rate. We compare the results of the stochastic fluid model with that of a simulation of the real system. We find the fluid model to not only be a close approximation, but also to exhibit the key qualitative properties of the original system. We conclude that stochastic fluid models show great potential in modeling a variety of content distribution systems.",
    keywords = "Fluid models, Palm calculus, Stochastic processes, Web caching",
    author = "Florence Cl{\'e}venot and Philippe Nain and Keith Ross",
    year = "2005",
    month = "1",
    doi = "10.1016/j.peva.2004.05.006",
    language = "English (US)",
    volume = "59",
    pages = "1--18",
    journal = "Performance Evaluation",
    issn = "0166-5316",
    publisher = "Elsevier",
    number = "1",

    }

    TY - JOUR

    T1 - Stochastic fluid models for cache clusters

    AU - Clévenot, Florence

    AU - Nain, Philippe

    AU - Ross, Keith

    PY - 2005/1

    Y1 - 2005/1

    N2 - Clusters of Web caches are extensively used by different types of organizations, including companies, universities, ISPs, and CDNs. To model Web caches, we must account for two types of stochastic events: objects being pulled into/out of the cache cluster at random times, and caches going up and down at random times. Detailed stochastic models of such complex systems quickly become intractable. In this paper we propose a stochastic fluid model which captures the salient characteristics of a cache cluster. The stochastic fluid model replaces the object arrivals to the cluster and departures (object modification/expiration) with a fluid flow, but maintains the up/down dynamics of the original system. The model can be applied to a variety of cluster routing policies, and provides a simple means to estimate the hit rate. We compare the results of the stochastic fluid model with that of a simulation of the real system. We find the fluid model to not only be a close approximation, but also to exhibit the key qualitative properties of the original system. We conclude that stochastic fluid models show great potential in modeling a variety of content distribution systems.

    AB - Clusters of Web caches are extensively used by different types of organizations, including companies, universities, ISPs, and CDNs. To model Web caches, we must account for two types of stochastic events: objects being pulled into/out of the cache cluster at random times, and caches going up and down at random times. Detailed stochastic models of such complex systems quickly become intractable. In this paper we propose a stochastic fluid model which captures the salient characteristics of a cache cluster. The stochastic fluid model replaces the object arrivals to the cluster and departures (object modification/expiration) with a fluid flow, but maintains the up/down dynamics of the original system. The model can be applied to a variety of cluster routing policies, and provides a simple means to estimate the hit rate. We compare the results of the stochastic fluid model with that of a simulation of the real system. We find the fluid model to not only be a close approximation, but also to exhibit the key qualitative properties of the original system. We conclude that stochastic fluid models show great potential in modeling a variety of content distribution systems.

    KW - Fluid models

    KW - Palm calculus

    KW - Stochastic processes

    KW - Web caching

    UR - http://www.scopus.com/inward/record.url?scp=9544219687&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=9544219687&partnerID=8YFLogxK

    U2 - 10.1016/j.peva.2004.05.006

    DO - 10.1016/j.peva.2004.05.006

    M3 - Article

    VL - 59

    SP - 1

    EP - 18

    JO - Performance Evaluation

    JF - Performance Evaluation

    SN - 0166-5316

    IS - 1

    ER -