[
https://issues.apache.org/jira/browse/CASSANDRA-4775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13852493#comment-13852493
]
Aleksey Yeschenko commented on CASSANDRA-4775:
----------------------------------------------
So, this thread has become quite overloaded. Will summarize it shortly in this
comment, and then move the actual work/discussion to CASSANDRA-6504.
The initial idea for the new design (a new cell for each increment/decrement,
then summing up on reads) and its variations didn't work out, for one reason or
another. The largest problems are the required coordination for collapsing the
increment history and difficulty in making it backward compatible with the
current implementation.
We decided to go for incremental improvements instead - namely, stop using
'local' shards altogether, and do explicit read-modify-write with just one
shard type ('global') instead. See
https://issues.apache.org/jira/browse/CASSANDRA-4775?focusedCommentId=13702042&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13702042
and the comments following it (plus
https://issues.apache.org/jira/browse/CASSANDRA-4071?focusedCommentId=13483381&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13483381).
This will fix, *at minimum*, the over counting issue with commit log replay,
CASSANDRA-4417, and CASSANDRA-4071, and, together with some related
improvements, drastically simplify counters code in general.
> Counters 2.0
> ------------
>
> Key: CASSANDRA-4775
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4775
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Arya Goudarzi
> Assignee: Aleksey Yeschenko
> Labels: counters
> Fix For: 2.1
>
>
> The existing partitioned counters remain a source of frustration for most
> users almost two years after being introduced. The remaining problems are
> inherent in the design, not something that can be fixed given enough
> time/eyeballs.
> Ideally a solution would give us
> - similar performance
> - less special cases in the code
> - potential for a retry mechanism
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
