Approximation-First Timeseries Monitoring Query At Scale

TL;DR

PromSketch introduces an approximation-first framework for timeseries monitoring queries that significantly reduces latency and operational costs by using sketch-based precomputation, effectively integrating with existing systems like Prometheus.

Contribution

It presents PromSketch, a novel approximation-based query framework that reduces latency and costs in timeseries monitoring systems by combining sketches and precomputation.

Findings

Achieves up to 100x reduction in query latency.

Reduces operational costs by over 100x.

Maintains at most 5% error in statistics.

Abstract

Timeseries monitoring systems such as Prometheus play a crucial role in gaining observability of the underlying system components. These systems collect timeseries metrics from various system components and perform monitoring queries over periodic window-based aggregations (i.e., rule queries). However, despite wide adoption, the operational costs and query latency of rule queries remain high. In this paper, we identify major bottlenecks associated with repeated data scans and query computations concerning window overlaps in rule queries, and present PromSketch, an approximation-first query framework as intermediate caches for monitoring systems. It enables low operational costs and query latency, by combining approximate window-based query frameworks and sketch-based precomputation. PromSketch is implemented as a standalone module that can be integrated into Prometheus and VictoriaMetrics, covering 70% of Prometheus' aggregation over time queries. Our evaluation shows that PromSketch achieves up to a two orders of magnitude reduction in query latency over Prometheus and VictoriaMetrics, while lowering operational dollar costs of query processing by two orders of magnitude compared to Prometheus and by at least 4x compared to VictoriaMetrics with at most 5% average errors across statistics. The source code has been made available at https://github.com/Froot-NetSys/promsketch.