If you want fast searching for some unique word or phrase across terabytes of logs (aka "needle in the haystack" type of search), then take a look at VictoriaLogs [1] (I'm the core developer). It uses bloom filters for quick skipping of data blocks, which do not contain the given word or phrase. Contrary to other open-source solutions for log storage and analysis, VictoriaLogs works efficiently with any types of logs containing any sets of fields, without the need in any configuration and tuning.

[1] https://docs.victoriametrics.com/victorialogs/