cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Nikolay (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-5479) hyper log log (hyperloglog) probabilistic counters
Date Tue, 16 Apr 2013 13:57:15 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-5479?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Nikolay updated CASSANDRA-5479:
-------------------------------

    Description: 
Would be nice if we can include probabilistic counters in Cassandra.
People do not use them because people do not know their power.

HyperLogLog also can merge counters, so in case of node disconnections,
will be easy to have different versions on each node. When node come online,
it just merge the data from other nodes.

Same can be used when adding data - similarly to Counter_CF,
HyperLogLog will need to read single replica and add there.

Adding can be even add without reading - just drop the new "item" somewhere,
but then reading will be much more slow.

  was:
Would be nice if we can include probabilistic counters in Cassandra.
People do not use them because people do not know their power.


    
> hyper log log (hyperloglog) probabilistic counters
> --------------------------------------------------
>
>                 Key: CASSANDRA-5479
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-5479
>             Project: Cassandra
>          Issue Type: Wish
>            Reporter: Nikolay
>            Priority: Minor
>
> Would be nice if we can include probabilistic counters in Cassandra.
> People do not use them because people do not know their power.
> HyperLogLog also can merge counters, so in case of node disconnections,
> will be easy to have different versions on each node. When node come online,
> it just merge the data from other nodes.
> Same can be used when adding data - similarly to Counter_CF,
> HyperLogLog will need to read single replica and add there.
> Adding can be even add without reading - just drop the new "item" somewhere,
> but then reading will be much more slow.

--
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

Mime
View raw message