[
https://issues.apache.org/jira/browse/CASSANDRA-3378?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13502305#comment-13502305
]
Michał Michalski commented on CASSANDRA-3378:
---------------------------------------------
Just pondering... In the existing code there's a BufferedOutputStream with
buffer size of 4096 bytes at first, which can be then "overwritten" by a stream
with default buffer size (no specified size is passed to constructor) if
compression is enabled.
My concern is: if there's a reason for setting different buffer size for
compressed and non-compressed connection by default, is this valid to apply the
same value (provided by user in config) for both connection types?
Shouldn't they be distinguished on a config level too? For example by
"splitting" storage_send_buff_size_bytes (that's how I temporarily named this
new param) into two options: storage_uncompressed_send_buff_size_bytes and
storage_compressed_send_buff_size_bytes?
Or maybe we assume that for this special environment (high latency connection)
the performance gain achieved by using larger buffer is big enough to just
ignore the case I mentioned and use the same value in these two places in code?
> Allow configuration of storage protocol socket buffer
> -----------------------------------------------------
>
> Key: CASSANDRA-3378
> URL: https://issues.apache.org/jira/browse/CASSANDRA-3378
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Brandon Williams
> Priority: Minor
> Labels: lhf
>
> Similar to rpc_[send,recv]_buff_size_in_bytes, we should expose this for high
> latency connections.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
