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.

Topical result caching in web search engines

“…In modern search engines, query processing represents one of the major performance bottlenecks, so caching can help to speed up the search engine performance as well as to reduce the latency perceived by the users. Caching can be applied at different granularity including query results [28], posting lists of query terms [39], and posting list intersections [31]. Saraiva et al [39] proposed a two-level architecture where the front-end machine caches the results of popular queries, while the back-end machines have a cache for the posting lists of most frequently requested terms.…”

A machine learning approach for result caching in web search engines

“…Altingovde, Ozcan, Cambazoglu, and Ulusoy (2011) coupled a result cache with a document id cache, which stores only document ids without snippets, further reducing the query traffic going to the backend search system. Long and Suel (2005) introduce, on top of result and posting list caching, a third level of cache where precomputed intersections of posting lists are stored. Li, Lee, Sivasubramaniam, and Giles (2007) propose a hybrid architecture involving result, posting list, and document caches.…”

Performance improvements for search systems using an integrated cache of lists + intersections

“…A further approach involves caching portions of a query (i.e., pairs of terms), as initially proposed in [14] and extended in [6]. This approach is named Intersection Caching and is implemented at search node level as well.…”