[
https://issues.apache.org/jira/browse/KAFKA-15418?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17761364#comment-17761364
]
Lucia Cerchie commented on KAFKA-15418:
---------------------------------------
suggested replacement text:
_Kafka supports this with an efficient batching format. A batch of messages can
be grouped together, compressed, and sent to the server in this form. The
broker decompresses the batch in order to validate it and potentially modify it
(e.g., if the topic is compacted, the broker will filter out records eligible
for compaction prior to writing to disk). This batch of messages is then
written to disk in compressed form. The batch will remain compressed in the log
and it will also be transmitted to the consumer in compressed form. The
consumer decompresses any compressed data that it receives._
> Update statement on decompression location
> -------------------------------------------
>
> Key: KAFKA-15418
> URL: https://issues.apache.org/jira/browse/KAFKA-15418
> Project: Kafka
> Issue Type: Improvement
> Components: docs
> Reporter: Lucia Cerchie
> Assignee: Lucia Cerchie
> Priority: Minor
> Labels: docs
> Fix For: 3.5.1
>
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> The docs [state|#L139-L140]]:
> "Kafka supports this with an efficient batching format. A batch of messages
> can be clumped together compressed and sent to the server in this form. This
> batch of messages will be written in compressed form and will
> remain compressed in the log and will only be decompressed by the
> consumer."
> However, brokers always perform some amount of batch decompression in order
> to validate data. Internally, the LogValidator class implements the
> validation logic, and message decompression ultimately happens in the
> relevant CompressionType's wrapForInput implementation. These lines need
> updating to reflect that, including the scenarios that require full payload
> decompression on the broker (compacted topic, or topic level compression
> codec different from producer's codec).
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
