Query Processing in Highly-Loaded Search Engines

While Web search engines are built to cope with a large number of queries, query traffic can exceed the maximum query rate supported by the underlying computing infrastructure. We study how response times and results vary when, in presence of high loads, some queries are either interrupted after a fixed time threshold elapses or dropped completely. Moreover, we introduce a novel dropping strategy, based on machine learned performance predictors to select the queries to drop in order to sustain the largest possible query rate with a relative degradation in effectiveness.