cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nathan McCall <n...@vervewireless.com>
Subject Re: get_slice() slow if more number of columns present in a SCF.
Date Wed, 03 Feb 2010 18:29:48 GMT
Alta,
The RowsCached attribute can take an integer, percentage or, like
KeysCachedFraction, a value between 0 and 1. I had a good idea already
how many records where going to be hot for the CF I was using to test,
show just matching that to available JVM memory was pretty straight
forward.

There was other stuff going on with the machine I was using to test
this with, so I would be interested to here from anybody else who had
dedicated hardware what sort of numbers they got over just relying on
a lot of ram and file system caching.

Cheers,
-Nate

On Wed, Feb 3, 2010 at 1:15 AM, envio user <enviouser@gmail.com> wrote:
> Hi Nate,
>
> On Wed, Feb 3, 2010 at 12:31 AM, Nathan McCall <nate@vervewireless.com> wrote:
>> Thank you for the benchmarks. What version of Cassandra are you using?
> I am using 0.5 release.
>> I had about 80% performance improvement on single node reads after
>> using a trunk build with the results from
>> https://issues.apache.org/jira/browse/CASSANDRA-688 (result caching)
>> and playing around with the configuration. I am not yet running this
>> in production though, so I cannot provide any real numbers.
>>
> Do you any specific config parameters needs to be used for this purpose?
>
>> That said, I have no intention of deploying a single node. I keep
>> seing performance concerns from folks on small or single node
>> clusters. My impression so far is that Cassandra may not be the right
>> solution for these types of deployments.
>
> As part of evaluation we are testing cassandra with single node. I
> agree for deployment it is not good having just single node.
>
> <<snip>>
>>
>> What I found to have been most valuable for my evaluation was to get a
>> good test together with some real data from our system and then add
>> nodes, remove nodes, break nodes, etc. and watch what happens. Once I
>> finish with this, it looks like I will have some solid numbers to do
>> some capacity planning for figuring out exactly how much hardware to
>> purchase and when I will need to add more.
> We are just doing some basic tests. If cassandra passes these tests,
> may be we will do tests you mentioned later.
>
> thanks,
> -Aita
>

Mime
View raw message