cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sylvain Lebresne (JIRA)" <>
Subject [jira] Commented: (CASSANDRA-1546) (Yet another) approach to counting
Date Mon, 27 Sep 2010 20:52:34 GMT


Sylvain Lebresne commented on CASSANDRA-1546:

Say at the beginning A has for K the value 4 (the row counter will have only one column, it's
'local counter 
column', that is a column whose name is A id). 
We add B. The sstable containing K with value 4 is streamed to B. For B, the column streamed
A is not it's 'local counter column' (it's the one of A), so it treats it as a normal column.

During the bootstrap, A receives update to K as a leader, say we add 2. A will insert 2 as

a LocalCounterColumn (terminology from the patch). Then A will read it's 'local counter column'.
This read will merge the old value 4, with the newly inserted update 2, and thus the result
the read is a column with value 6. Then A will replicate the update to B by sending to B the
value 6 (more precisely a column whose name is A id, value is 6 and timestamp is the timestamp
of the new update. B receives that and treat that as a standard column (it will overwrite
the streamed
column with the value 4 because the timestamp will be greater).

> (Yet another) approach to counting
> ----------------------------------
>                 Key: CASSANDRA-1546
>                 URL:
>             Project: Cassandra
>          Issue Type: New Feature
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>             Fix For: 0.7.0
>         Attachments: 0001-Remove-IClock-from-internals.patch, 0002-Counters.patch, 0003-Generated-thrift-files-changes.patch
> This could be described as a mix between CASSANDRA-1072 without clocks and CASSANDRA-1421.
> More details in the comment below.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message