You are viewing a plain text version of this content. The canonical link for it is here.
Posted to rampart-dev@ws.apache.org by Rajan Gupta <Ra...@Yodlee.com> on 2009/11/19 03:12:46 UTC
Rampart performance
Hi,
Is there any document about the performance benchmarks of Rampart
1.4? I am using Rampart 1.4 with axis2 version 1.4.1 with full body
signing and encryption. The performance degrades a lot when we enabled
WSSE in our web services. Below is what we are seeing in our initial
benchmarking:
TPS CPU Utilization
Without WSSE 7.1 75%
With WSSE 32 50%
The server with WSSE gives one-fourth the TPS with 50% more CPU
utilization.
I am trying to find if there is anyone else who has done similar
performance benchmarking of Rampart.
Thanks in advance,
Rajan
Re: Rampart performance
Posted by Prabath Siriwardena <pr...@wso2.com>.
Rajan Gupta wrote:
> We were expecting some performance degradation but the performance
> degradation that we are seeing is huge. And if I get you correctly, the
> kind of performance degradation we saw is more or less expected with
> Rampart version 1.4.
>
Rampart 1.5 to be released or the trunk will better in performance with
respect to Rampart 1.4 - specially with fix Thilina did to cache keys.
If you are to rerun the test against it - please make sure you follow
the steps in [1].
Thanks & regards.
-Prabath
[1]:http://marc.info/?l=rampart-dev&m=125622929630598&w=2
>
> Thanks,
> Rajan
>
> -----Original Message-----
> From: Prabath Siriwardena [mailto:prabath@wso2.com]
> Sent: Wednesday, November 18, 2009 6:30 PM
> To: rampart-dev@ws.apache.org
> Subject: Re: Rampart performance
>
> Rajan Gupta wrote:
>
>> The server with WSSE gives one-fourth the TPS with 50% more CPU
>> utilization.
>>
> Adding WSSE has a performance hit against sending in clear text - as
> well as over SSL. The TPS value comparison you ended up with, seems to
> be correct [although in the table it seems to be presented other way
> around]
>
> We have also done certain performance improvements[1] in Rampart trunk
> [and also will be available with Rampart 1.5] - so you may rerun your
> tests against the trunk as well.
>
> Thanks & regards.
> -Prabath
>
> [1]:http://marc.info/?l=rampart-dev&m=125622929630598&w=2
>
>>
>>
>>
>>
>> I am trying to find if there is anyone else who has done similar
>> performance benchmarking of Rampart.
>>
>>
>>
>> Thanks in advance,
>>
>> Rajan
>>
>>
>>
>>
>
>
>
RE: Rampart performance
Posted by Rajan Gupta <Ra...@Yodlee.com>.
Thanks Prabhat for the reply. Yes, sorry about the typos. The numbers
were presented the other way around. They should have been:
TPS CPU
With WSSE 7.1 75%
Without WSSE 32 50%
We were expecting some performance degradation but the performance
degradation that we are seeing is huge. And if I get you correctly, the
kind of performance degradation we saw is more or less expected with
Rampart version 1.4.
Thanks,
Rajan
-----Original Message-----
From: Prabath Siriwardena [mailto:prabath@wso2.com]
Sent: Wednesday, November 18, 2009 6:30 PM
To: rampart-dev@ws.apache.org
Subject: Re: Rampart performance
Rajan Gupta wrote:
> The server with WSSE gives one-fourth the TPS with 50% more CPU
> utilization.
Adding WSSE has a performance hit against sending in clear text - as
well as over SSL. The TPS value comparison you ended up with, seems to
be correct [although in the table it seems to be presented other way
around]
We have also done certain performance improvements[1] in Rampart trunk
[and also will be available with Rampart 1.5] - so you may rerun your
tests against the trunk as well.
Thanks & regards.
-Prabath
[1]:http://marc.info/?l=rampart-dev&m=125622929630598&w=2
>
>
>
>
> I am trying to find if there is anyone else who has done similar
> performance benchmarking of Rampart.
>
>
>
> Thanks in advance,
>
> Rajan
>
>
>
Re: Rampart performance
Posted by Prabath Siriwardena <pr...@wso2.com>.
Rajan Gupta wrote:
> The server with WSSE gives one-fourth the TPS with 50% more CPU
> utilization.
Adding WSSE has a performance hit against sending in clear text - as
well as over SSL. The TPS value comparison you ended up with, seems to
be correct [although in the table it seems to be presented other way around]
We have also done certain performance improvements[1] in Rampart trunk
[and also will be available with Rampart 1.5] - so you may rerun your
tests against the trunk as well.
Thanks & regards.
-Prabath
[1]:http://marc.info/?l=rampart-dev&m=125622929630598&w=2
>
>
>
>
> I am trying to find if there is anyone else who has done similar
> performance benchmarking of Rampart.
>
>
>
> Thanks in advance,
>
> Rajan
>
>
>