@gaodayue thank you for the details! Probably we can optimize the initial dictionary size based on 1) spilling is enabled, 2) the column is dictionary encoded, 3) the cardinality of the column, and 4) the filter selectivity.
[ Full content available at: https://github.com/apache/incubator-druid/issues/6255 ] This message was relayed via gitbox.apache.org for [email protected]
