You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by Kent Tong <fr...@yahoo.com> on 2011/12/15 11:58:33 UTC

performance reaching plateau while the hardware is still idle

Hi,

I am running a performance test for Cassandra 1.0.5. It can perform about 1500 business operation (one read+one write to the same row) per second. However, the CPU is still 85% idle (as shown by vmstat) and the IO utilization is less than a few percent (as shown by iostat). "nodetool tpstats" shows basically no active and pending threads. I can run several such test clients concurrently, achieving the same operations per second without increasing the hardware utilization. So, why the performance has reached a plateau while there is still idle hardware resources?


Thanks in advance for any idea!

Re: performance reaching plateau while the hardware is still idle

Posted by Kent Tong <fr...@yahoo.com>.
Dear all,

Thanks for the good suggestions! I believe it is because the test client is single-threaded, so only server thread is serving it, so the rest of the cores are just sitting idle.

Thanks!
 
--
Author of books for learning CXF, Axis2, Wicket, JSF (http://www.agileskills2.org)


________________________________
 From: Peter Tillotson <sl...@yahoo.co.uk>
To: "user@cassandra.apache.org" <us...@cassandra.apache.org>; Kent Tong <fr...@yahoo.com> 
Sent: Friday, December 16, 2011 1:45 AM
Subject: Re: performance reaching plateau while the hardware is still idle
 
May I suggest dstat, does cpu, memory, and io on one console
dstat -vn 3

So possible causes
 * Not sufficient parrallelism in the client
 * Server has too few threads
 * You are not CPU bound, network or disk may be the bottleneck 

p


----- Original Message -----
From: Kent Tong <fr...@yahoo.com>
To: "user@cassandra.apache.org" <us...@cassandra.apache.org>
Cc: 
Sent: Thursday, 15 December 2011, 10:58
Subject: performance reaching plateau while the hardware is still idle

Hi,

I am running a performance test for Cassandra 1.0.5. It can perform about 1500 business operation (one read+one write to the same row) per second. However, the CPU is still 85% idle (as shown by vmstat) and the IO utilization is less than a few percent (as shown by iostat). "nodetool tpstats" shows basically no active and pending threads. I can run several such test clients concurrently, achieving the same operations per second without increasing the hardware utilization. So, why the performance has reached a plateau while there is still idle hardware resources?


Thanks in advance for any idea!

Re: performance reaching plateau while the hardware is still idle

Posted by Peter Tillotson <sl...@yahoo.co.uk>.
May I suggest dstat, does cpu, memory, and io on one console
dstat -vn 3

So possible causes
 * Not sufficient parrallelism in the client
 * Server has too few threads
 * You are not CPU bound, network or disk may be the bottleneck 

p


----- Original Message -----
From: Kent Tong <fr...@yahoo.com>
To: "user@cassandra.apache.org" <us...@cassandra.apache.org>
Cc: 
Sent: Thursday, 15 December 2011, 10:58
Subject: performance reaching plateau while the hardware is still idle

Hi,

I am running a performance test for Cassandra 1.0.5. It can perform about 1500 business operation (one read+one write to the same row) per second. However, the CPU is still 85% idle (as shown by vmstat) and the IO utilization is less than a few percent (as shown by iostat). "nodetool tpstats" shows basically no active and pending threads. I can run several such test clients concurrently, achieving the same operations per second without increasing the hardware utilization. So, why the performance has reached a plateau while there is still idle hardware resources?


Thanks in advance for any idea!

Re: performance reaching plateau while the hardware is still idle

Posted by ruslan usifov <ru...@gmail.com>.
Use parallel test:-)))

2011/12/15 Kent Tong <fr...@yahoo.com>

> Hi,
>
> I am running a performance test for Cassandra 1.0.5. It can perform about
> 1500 business operation (one read+one write to the same row) per second.
> However, the CPU is still 85% idle (as shown by vmstat) and the IO
> utilization is less than a few percent (as shown by iostat). "nodetool
> tpstats" shows basically no active and pending threads. I can run several
> such test clients concurrently, achieving the same operations per second
> without increasing the hardware utilization. So, why the performance has
> reached a plateau while there is still idle hardware resources?
>
>
> Thanks in advance for any idea!
>