[
https://issues.apache.org/jira/browse/CASSANDRA-15657?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17074266#comment-17074266
]
Dinesh Joshi edited comment on CASSANDRA-15657 at 4/3/20, 4:26 AM:
-------------------------------------------------------------------
I did consider this when I implemented Zero Copy Streaming check. However after
consulting [~marcuse] and [~aleksey], I determined that the only way to
_correctly_ to implement the containment check is to enumerate all tokens. As
to why it was only enabled for LCS and not other compaction strategies, it came
down to the cost of running this check. Turns out this check can turn out to
be quite expensive with STCS while the LCS has a roughly fixed cost associated
with it. To follow up I filed CASSANDRA-14586.
Overall, I'm -1 on moving ahead with this patch unless there is something we
are missing here.
was (Author: djoshi3):
I did consider this when I implemented Zero Copy Streaming check. However after
consulting [~marcuse] and [~aleksey], I determined the only way to _correctly_
determine it is to enumerate all tokens. As to why it was only enabled for LCS
and not other compaction strategies, it came down to the cost of running this
check. Turns out this check can turn out to be quite expensive with STCS while
the LCS has a roughly fixed cost associated with it. To follow up I filed
CASSANDRA-14586.
Overall, I'm -1 on moving ahead with this patch unless there is something we
are missing here.
> Improve zero-copy-streaming containment check by using file sections
> --------------------------------------------------------------------
>
> Key: CASSANDRA-15657
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15657
> Project: Cassandra
> Issue Type: Improvement
> Components: Legacy/Streaming and Messaging
> Reporter: ZhaoYang
> Assignee: ZhaoYang
> Priority: Normal
> Fix For: 4.0
>
>
> Currently zero copy streaming is only enabled for leveled-compaction strategy
> and it checks if all keys in the sstables are included in the transferred
> ranges.
> This is very inefficient. The containment check can be improved by checking
> if transferred sections (the transferred file positions) cover entire sstable.
> I also enabled ZCS for all compaction strategies since the new containment
> check is very fast..
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
