Re: Apache Cassandra 2.0.0 rc1

2013-08-12 Thread Radim Kolar 

thrift, sync

Re: Apache Cassandra 2.0.0 rc1

2013-08-12 Thread Michael Kjellman 
Should probably use this instead. All benchmarks show it¹s faster than
anything that has ever been written before.

https://issues.apache.org/jira/browse/CASSANDRA-5582

On 8/12/13, 12:47 AM, Radim Kolar h...@filez.com wrote:

thrift, sync

Re: Apache Cassandra 2.0.0 rc1

2013-08-12 Thread Jonathan Ellis 
We haven't seen that in our tests.  Suggest bisecting to figure out
what changed it for your workload.

On Mon, Aug 12, 2013 at 2:47 AM, Radim Kolar h...@filez.com wrote:
 thrift, sync



-- 
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder, http://www.datastax.com
@spyced

Re: Apache Cassandra 2.0.0 rc1

2013-08-12 Thread Ryan McGuire 
I've seen similar slowdowns in 2.0 - roughly the same 15%.

For instance, http://i.imgur.com/Qi6oo62.png

That's a comparison of 1.2.6 to 2.0beta1. Ignore the lines that say
speculative_retry=95percentile and ALWAYS.
Default 1.2 beats Default 2.0 in overall operation time.

Bisecting the cause of the problem is on my TODO list.

On Mon, Aug 12, 2013 at 2:27 PM, Jonathan Ellis jbel...@gmail.com wrote:

 We haven't seen that in our tests.  Suggest bisecting to figure out
 what changed it for your workload.

 On Mon, Aug 12, 2013 at 2:47 AM, Radim Kolar h...@filez.com wrote:
  thrift, sync



 --
 Jonathan Ellis
 Project Chair, Apache Cassandra
 co-founder, http://www.datastax.com
 @spyced

Re: Apache Cassandra 2.0.0 rc1

2013-08-11 Thread Radim Kolar 

Dne 10.8.2013 21:30, Brandon Williams napsal(a):

Make a conf/triggers directory and that will fix it. We fixed this in trunk 
already.

yes, that fixed it.

2.0 is considerably slower then 1.2 for cpu bound tasks, average 
throughput is -15% at 50 threads.

2.0 with 20 threads burst thruput with is just 1/2 of 1.2 at 50 threads.

Re: Apache Cassandra 2.0.0 rc1

2013-08-11 Thread Michael Kjellman 
Thrift? Native?

If Thrift- what threading model?

 On Aug 11, 2013, at 4:13 PM, Radim Kolar h...@filez.com wrote:
 
 Dne 10.8.2013 21:30, Brandon Williams napsal(a):
 Make a conf/triggers directory and that will fix it. We fixed this in trunk 
 already.
 yes, that fixed it.
 
 2.0 is considerably slower then 1.2 for cpu bound tasks, average 
 throughput is -15% at 50 threads.
 2.0 with 20 threads burst thruput with is just 1/2 of 1.2 at 50 threads.