Three-level caching for efficient query processing in large Web search engines

Xiaohui Long, Torsten Suel

    Research output: Contribution to journalArticle

    Abstract

    Large web search engines have to answer thousands of queries per second with interactive response times. Due to the sizes of the data sets involved, often in the range of multiple terabytes, a single query may require the processing of hundreds of megabytes or more of index data. To keep up with this immense workload, large search engines employ clusters of hundreds or thousands of machines, and a number of techniques such as caching, index compression, and index and query pruning are used to improve scalability. In particular, two-level caching techniques cache results of repeated identical queries at the frontend, while index data for frequently used query terms are cached in each node at a lower level. We propose and evaluate a three-level caching scheme that adds an intermediate level of caching for additional performance gains. This intermediate level attempts to exploit frequently occurring pairs of terms by caching intersections or projections of the corresponding inverted lists. We propose and study several offline and online algorithms for the resulting weighted caching problem, which turns out to be surprisingly rich in structure. Our experimental evaluation based on a large web crawl and real search engine query log shows significant performance gains for the best schemes, both in isolation and in combination with the other caching levels. We also observe that a careful selection of cache admission and eviction policies is crucial for best overall performance.

    Original languageEnglish (US)
    Pages (from-to)369-395
    Number of pages27
    JournalWorld Wide Web
    Volume9
    Issue number4
    DOIs
    StatePublished - Dec 2006

    Fingerprint

    Query processing
    Search engines
    World Wide Web
    Scalability
    Processing

    Keywords

    • Caching
    • Inverted index
    • Search engine architecture
    • Search engine query processing
    • Web search

    ASJC Scopus subject areas

    • Computer Networks and Communications

    Cite this

    Three-level caching for efficient query processing in large Web search engines. / Long, Xiaohui; Suel, Torsten.

    In: World Wide Web, Vol. 9, No. 4, 12.2006, p. 369-395.

    Research output: Contribution to journalArticle

    Long, Xiaohui ; Suel, Torsten. / Three-level caching for efficient query processing in large Web search engines. In: World Wide Web. 2006 ; Vol. 9, No. 4. pp. 369-395.
    @article{87c21de3d11a48dcb950bf201b811edd,
    title = "Three-level caching for efficient query processing in large Web search engines",
    abstract = "Large web search engines have to answer thousands of queries per second with interactive response times. Due to the sizes of the data sets involved, often in the range of multiple terabytes, a single query may require the processing of hundreds of megabytes or more of index data. To keep up with this immense workload, large search engines employ clusters of hundreds or thousands of machines, and a number of techniques such as caching, index compression, and index and query pruning are used to improve scalability. In particular, two-level caching techniques cache results of repeated identical queries at the frontend, while index data for frequently used query terms are cached in each node at a lower level. We propose and evaluate a three-level caching scheme that adds an intermediate level of caching for additional performance gains. This intermediate level attempts to exploit frequently occurring pairs of terms by caching intersections or projections of the corresponding inverted lists. We propose and study several offline and online algorithms for the resulting weighted caching problem, which turns out to be surprisingly rich in structure. Our experimental evaluation based on a large web crawl and real search engine query log shows significant performance gains for the best schemes, both in isolation and in combination with the other caching levels. We also observe that a careful selection of cache admission and eviction policies is crucial for best overall performance.",
    keywords = "Caching, Inverted index, Search engine architecture, Search engine query processing, Web search",
    author = "Xiaohui Long and Torsten Suel",
    year = "2006",
    month = "12",
    doi = "10.1007/s11280-006-0221-0",
    language = "English (US)",
    volume = "9",
    pages = "369--395",
    journal = "World Wide Web",
    issn = "1386-145X",
    publisher = "Springer New York",
    number = "4",

    }

    TY - JOUR

    T1 - Three-level caching for efficient query processing in large Web search engines

    AU - Long, Xiaohui

    AU - Suel, Torsten

    PY - 2006/12

    Y1 - 2006/12

    N2 - Large web search engines have to answer thousands of queries per second with interactive response times. Due to the sizes of the data sets involved, often in the range of multiple terabytes, a single query may require the processing of hundreds of megabytes or more of index data. To keep up with this immense workload, large search engines employ clusters of hundreds or thousands of machines, and a number of techniques such as caching, index compression, and index and query pruning are used to improve scalability. In particular, two-level caching techniques cache results of repeated identical queries at the frontend, while index data for frequently used query terms are cached in each node at a lower level. We propose and evaluate a three-level caching scheme that adds an intermediate level of caching for additional performance gains. This intermediate level attempts to exploit frequently occurring pairs of terms by caching intersections or projections of the corresponding inverted lists. We propose and study several offline and online algorithms for the resulting weighted caching problem, which turns out to be surprisingly rich in structure. Our experimental evaluation based on a large web crawl and real search engine query log shows significant performance gains for the best schemes, both in isolation and in combination with the other caching levels. We also observe that a careful selection of cache admission and eviction policies is crucial for best overall performance.

    AB - Large web search engines have to answer thousands of queries per second with interactive response times. Due to the sizes of the data sets involved, often in the range of multiple terabytes, a single query may require the processing of hundreds of megabytes or more of index data. To keep up with this immense workload, large search engines employ clusters of hundreds or thousands of machines, and a number of techniques such as caching, index compression, and index and query pruning are used to improve scalability. In particular, two-level caching techniques cache results of repeated identical queries at the frontend, while index data for frequently used query terms are cached in each node at a lower level. We propose and evaluate a three-level caching scheme that adds an intermediate level of caching for additional performance gains. This intermediate level attempts to exploit frequently occurring pairs of terms by caching intersections or projections of the corresponding inverted lists. We propose and study several offline and online algorithms for the resulting weighted caching problem, which turns out to be surprisingly rich in structure. Our experimental evaluation based on a large web crawl and real search engine query log shows significant performance gains for the best schemes, both in isolation and in combination with the other caching levels. We also observe that a careful selection of cache admission and eviction policies is crucial for best overall performance.

    KW - Caching

    KW - Inverted index

    KW - Search engine architecture

    KW - Search engine query processing

    KW - Web search

    UR - http://www.scopus.com/inward/record.url?scp=33846307437&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=33846307437&partnerID=8YFLogxK

    U2 - 10.1007/s11280-006-0221-0

    DO - 10.1007/s11280-006-0221-0

    M3 - Article

    AN - SCOPUS:33846307437

    VL - 9

    SP - 369

    EP - 395

    JO - World Wide Web

    JF - World Wide Web

    SN - 1386-145X

    IS - 4

    ER -