Abstract

In interactive services such as web search, recommendations, games and finance, reducing the tail latency is crucial to provide fast response to every user. Using web search as a driving example, we systematically characterize interactive workload to identify the opportunities and challenges for reducing tail latency. We find that the workload consists of mainly short requests that do not benefit from parallelism, and a few long requests which significantly impact the tail but exhibit high parallelism speedup. This motivates estimating request execution time, using a predictor, to identify long requests and to parallelize them. Prediction, however, is not perfect; a long request mispredicted as short is likely to contribute to the server tail latency, setting a ceiling on the achievable tail latency. We propose TPC, an approach that combines prediction information judiciously with dynamic correction for inaccurate prediction. Dynamic correction increases parallelism to accelerate a long request that is mispredicted as short. TPC carefully selects the appropriate target latencies based on system load and parallelism efficiency to reduce tail latency.

We implement TPC and several prior approaches to compare them experimentally on a single search server and on a cluster of 40 search servers. The experimental results show that TPC reduces the 99th- and 99.9th-percentile latency by up to 40% compared with the best prior work. Moreover, we evaluate TPC on a finance server, demonstrating its effectiveness on reducing tail latency of interactive services beyond web search.