cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Philippe <watche...@gmail.com>
Subject Re: Where is the Overview Documentation on Counters?
Date Fri, 10 Jun 2011 22:25:09 GMT
Ian,
Have you been able to measure the performance penalty of running at CL=ALL ?
Right now I'm spreading updates over such counter columns across workers so
they don't overlap keys and that way I don't go to CL=ALL but maybe that's
not worth it? Any input?

Thanks
Philippe

2011/6/10 Ian Holsman <hadoop@holsman.net>

> Hi AJ.
>
> Counters are really cool for certain things..
>
> The main benefit (from a high level perspective) is that you don't have to
> read the record in to find the old value. (and stick a lock on the record to
> prevent it from changing underneath you).
>
> what I use them for is to increment page-views. I just read a log-line in,
> and can update the rows without having to read it beforehand.. which is nice
> performance wise.
>
> On Jun 10, 2011, at 2:18 PM, AJ wrote:
>
> > I can't find any that gives an overview of their purpose/benefits/etc,
> only how to code them.  I can only guess that they are more efficient for
> some reason but don't know exactly why or exactly what conditions I would
> choose to use them over a regular column.
> >
> > Thanks!
>
>

Mime
View raw message