Chia-Ping Tsai created KAFKA-21081:
--------------------------------------
Summary: Don't allocate key and value if all we care about is the
timestamp
Key: KAFKA-21081
URL: https://issues.apache.org/jira/browse/KAFKA-21081
Project: Kafka
Issue Type: Improvement
Reporter: Chia-Ping Tsai
Assignee: Chia-Ping Tsai
{{FileRecords.searchForTimestamp}}
{{RemoteLogManager.lookupTimestamp}}
{{RecordBatch.offsetOfMaxTimestamp}}
{{}}
Those paths don't need the key and value, so using {{skipKeyValueIterator}} is
the better choice. However, {{skipKeyValueIterator}} is declared on
{{MutableRecordBatch}} only, and {{FileChannelRecordBatch}} doesn't implement
it.
{{}}
{{In short, we should move }}{{skipKeyValueIterator}}{{ to
}}{{{}RecordBatch{}}}{{{}, make {}}}{{FileChannelRecordBatch}}{{ support it by
delegating to }}{{{}loadFullBatch(){}}}{{{}, and replace
{}}}{{streamingIterator}}{{ with }}{{skipKeyValueIterator}}{{ on those paths 🙂}}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)