Company
Date Published
Author
-
Word count
681
Language
-
Hacker News points
None

Summary

Elasticsearch 7.14 introduces the match_only_text field type, designed to reduce disk space usage in logging datasets by approximately 10% compared to the traditional text field type. This new field type is particularly beneficial for logging use cases, as it indexes only a subset of the information, thereby decreasing CPU and disk space requirements without sacrificing the ability to query data interactively. Although relevancy scores and span queries are not supported, and phrase and interval queries run slower than on text fields, other queries perform equally or even slightly faster. The release follows a recent trend of index size reductions, including a similar 10% decrease in version 7.10 through improved stored fields compression. By upgrading to the latest version of the Elastic Stack, users can leverage these space-saving benefits, which are achieved by eliminating the indexing of length normalization factors, term frequencies, and positions that are not crucial for log analysis. Match_only_text also employs runtime field concepts to handle phrase queries efficiently when necessary, ensuring better performance than a linear scan.