You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@etch.apache.org by Siva Kumar <si...@gmail.com> on 2013/04/01 13:00:42 UTC
Etch Performance Details
Hi All,
I have seen the performance details of Etch in the documentation. I
could see the following numbers in the documentation.
*System* *Calls/Second*Etch Java 13.3k Thrift Java13.5k Java RMI7.4k CORBA
Java2.4k JAX-WS943 Apache Axis2206
1. Could please clarify the hardware details,OS details about this test?
2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking
method call?
I am planing to use Etch to my project. Please suggest how to validate the
performance.We need to validate the performance before deciding anything on
the next step..Please reply as early as possible.
Regards,
Siva
Re: Etch Performance Details
Posted by scott comer <we...@mac.com>.
On 4/2/2013 10:44 AM, Yann Nicolas wrote:
>
> Thanks for your reply,
>
> I consider that reaching 5K in one server is good enough if I can
> balance the load through several active nodes using a Network Load
> Balancer. You know if the TCP selector transport is stable for C#?
>
for 5k, maybe. dunno. you'll have to fire it up and try it.
but you'll have to tune windows to get to 5k, even, and be sure your
traffic generator is on a different machine. and you have to be careful
to model how users actually arrive at the system (not all at once).
making or breaking connections is tricky, so model the connection
process on how users actually work. (connect, request, pause, request,
pause, request, pause, ..., disconnect) then repeat. throw in some
randomness. your users scripts should include metrics (performance
guaranty, reliability).
systems often have a busy hour, and a defined rate of service required
(100,000 users arriving within 5 minutes around 7 pm). make sure you
simulate that rate and still have some headroom left (try not to commit
your system resources more than 50% to leave room for bursts and
growth). make sure when you load the system fully it continues to
function without thrashing.
scott out
> Regards,
>
> Yann
>
> Le 2 avr. 2013 09:34, "Scott Comer" <wert1y@mac.com
> <ma...@mac.com>> a écrit :
>
> That one is harder. I got up to 30,000 one time. Goal was 50,000
> but ran out of memory. Os tuning is as critical as is choice of
> os. Hardware resources, memory and processors also critical. Not
> easy, there is scarce info on how to do it. The selector based
> transport, tcp2, is required.
>
> Theoretically, you could go as high as 150,000 (kernel limits).
>
> Sent from my iPhone
>
> On Apr 2, 2013, at 10:11 AM, Yann Nicolas <yannart@gmail.com
> <ma...@gmail.com>> wrote:
>
>> Hi,
>>
>> Somebody have measured the maximum concurrent connections we can
>> have (clients connected simultaneously to the server? In a
>> Windows Server 2008 with a C# server for example. Which is
>> usually the bottleneck? The number of connected sockets? The RAM
>> used by the process? CPU?
>>
>> Thanks,
>>
>> Yann
>>
>> Le 2 avr. 2013 08:46, "Scott Comer" <wert1y@mac.com
>> <ma...@mac.com>> a écrit :
>>
>> The hardware for those tests that were run years ago is now
>> meaningless. Pentium 4. I've run those tests last year with
>> very different results. The only thing interesting there is
>> the relative performance compared to other tech, which hasn't
>> changed a lot.
>>
>> What matters and nobody else can answer this for you, ever,
>> is how your app performs in your environment.
>>
>> By all means, run the tests. Modify them to more closely
>> model your app and match your needs. As previously stated,
>> see examples/Perf.
>>
>> Scott out
>>
>> Sent from my iPhone
>>
>> On Apr 2, 2013, at 8:37 AM, Siva Kumar
>> <sivakumarbe24@gmail.com <ma...@gmail.com>> wrote:
>>
>>> Thanks for the reply Martin..
>>>
>>> Scott ,
>>> Please share if you any information on this,
>>>
>>> Regards,
>>> Siva
>>>
>>>
>>> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith
>>> <Martin.Veith@bmw-carit.de
>>> <ma...@bmw-carit.de>> wrote:
>>>
>>> Hi Siva,
>>>
>>> > 1. Could please clarify the hardware details,OS
>>> details about this test?
>>> the performance measurements shown in the documentation
>>> have been done before I joined the team, so
>>> unfortunately I cannot tell you some more details on the
>>> hardware or OS used.
>>> @Scott: Maybe you can help out?
>>>
>>> > 2. Is this asynchronous/ Non Blocking method call or
>>> synchronous /Blocking method call?
>>> The performance test itself is available in the
>>> example/perf directory, sou you could run them by your
>>> own on your specific hardware/OS.
>>> The IDL description [1] shows the interface (e.g. the
>>> "add" function mentioned in the documentation), the
>>> implementation is available in [2]. For "add" there are
>>> reports for both synchronous and asynchronous calls.
>>>
>>>
>>> Best regards,
>>> Martin
>>>
>>> [1]
>>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>>> [2]
>>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>>
>>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com
>>> <ma...@gmail.com>]
>>> Sent: Monday, April 01, 2013 1:01 PM
>>> To: user@etch.apache.org <ma...@etch.apache.org>
>>> Subject: Etch Performance Details
>>>
>>>
>>>
>>> Hi All,
>>>
>>> I have seen the performance details of Etch in the
>>> documentation. I could see the following numbers in the
>>> documentation.
>>>
>>> System
>>> Calls/Second
>>> Etch Java
>>> 13.3k
>>> Thrift Java
>>> 13.5k
>>> Java RMI
>>> 7.4k
>>> CORBA Java
>>> 2.4k
>>> JAX-WS
>>> 943
>>> Apache Axis2
>>> 206
>>>
>>> 1. Could please clarify the hardware details,OS details
>>> about this test?
>>> 2. Is this asynchronous/ Non Blocking method call or
>>> synchronous /Blocking method call?
>>>
>>> I am planing to use Etch to my project. Please suggest
>>> how to validate the performance.We need to validate the
>>> performance before deciding anything on the next
>>> step..Please reply as early as possible.
>>>
>>> Regards,
>>> Siva
>>>
>>>
Re: Etch Performance Details
Posted by Yann Nicolas <ya...@gmail.com>.
Thanks for your reply,
I consider that reaching 5K in one server is good enough if I can balance
the load through several active nodes using a Network Load Balancer. You
know if the TCP selector transport is stable for C#?
Regards,
Yann
Le 2 avr. 2013 09:34, "Scott Comer" <we...@mac.com> a écrit :
> That one is harder. I got up to 30,000 one time. Goal was 50,000 but ran
> out of memory. Os tuning is as critical as is choice of os. Hardware
> resources, memory and processors also critical. Not easy, there is scarce
> info on how to do it. The selector based transport, tcp2, is required.
>
> Theoretically, you could go as high as 150,000 (kernel limits).
>
> Sent from my iPhone
>
> On Apr 2, 2013, at 10:11 AM, Yann Nicolas <ya...@gmail.com> wrote:
>
> Hi,
>
> Somebody have measured the maximum concurrent connections we can have
> (clients connected simultaneously to the server? In a Windows Server 2008
> with a C# server for example. Which is usually the bottleneck? The number
> of connected sockets? The RAM used by the process? CPU?
>
> Thanks,
>
> Yann
> Le 2 avr. 2013 08:46, "Scott Comer" <we...@mac.com> a écrit :
>
>> The hardware for those tests that were run years ago is now meaningless.
>> Pentium 4. I've run those tests last year with very different results. The
>> only thing interesting there is the relative performance compared to other
>> tech, which hasn't changed a lot.
>>
>> What matters and nobody else can answer this for you, ever, is how your
>> app performs in your environment.
>>
>> By all means, run the tests. Modify them to more closely model your app
>> and match your needs. As previously stated, see examples/Perf.
>>
>> Scott out
>>
>> Sent from my iPhone
>>
>> On Apr 2, 2013, at 8:37 AM, Siva Kumar <si...@gmail.com> wrote:
>>
>> Thanks for the reply Martin..
>>
>> Scott ,
>> Please share if you any information on this,
>>
>> Regards,
>> Siva
>>
>>
>> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de>wrote:
>>
>>> Hi Siva,
>>>
>>> > 1. Could please clarify the hardware details,OS details about this
>>> test?
>>> the performance measurements shown in the documentation have been done
>>> before I joined the team, so unfortunately I cannot tell you some more
>>> details on the hardware or OS used.
>>> @Scott: Maybe you can help out?
>>>
>>> > 2. Is this asynchronous/ Non Blocking method call or synchronous
>>> /Blocking method call?
>>> The performance test itself is available in the example/perf directory,
>>> sou you could run them by your own on your specific hardware/OS.
>>> The IDL description [1] shows the interface (e.g. the "add" function
>>> mentioned in the documentation), the implementation is available in [2].
>>> For "add" there are reports for both synchronous and asynchronous calls.
>>>
>>>
>>> Best regards,
>>> Martin
>>>
>>> [1]
>>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>>> [2]
>>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>>
>>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
>>> Sent: Monday, April 01, 2013 1:01 PM
>>> To: user@etch.apache.org
>>> Subject: Etch Performance Details
>>>
>>>
>>>
>>> Hi All,
>>>
>>> I have seen the performance details of Etch in the documentation. I
>>> could see the following numbers in the documentation.
>>>
>>> System
>>> Calls/Second
>>> Etch Java
>>> 13.3k
>>> Thrift Java
>>> 13.5k
>>> Java RMI
>>> 7.4k
>>> CORBA Java
>>> 2.4k
>>> JAX-WS
>>> 943
>>> Apache Axis2
>>> 206
>>>
>>> 1. Could please clarify the hardware details,OS details about this test?
>>> 2. Is this asynchronous/ Non Blocking method call or
>>> synchronous /Blocking method call?
>>>
>>> I am planing to use Etch to my project. Please suggest how to validate
>>> the performance.We need to validate the performance before deciding
>>> anything on the next step..Please reply as early as possible.
>>>
>>> Regards,
>>> Siva
>>>
>>>
>>
Re: Etch Performance Details
Posted by Scott Comer <we...@mac.com>.
That one is harder. I got up to 30,000 one time. Goal was 50,000 but ran out of memory. Os tuning is as critical as is choice of os. Hardware resources, memory and processors also critical. Not easy, there is scarce info on how to do it. The selector based transport, tcp2, is required.
Theoretically, you could go as high as 150,000 (kernel limits).
Sent from my iPhone
On Apr 2, 2013, at 10:11 AM, Yann Nicolas <ya...@gmail.com> wrote:
> Hi,
>
> Somebody have measured the maximum concurrent connections we can have (clients connected simultaneously to the server? In a Windows Server 2008 with a C# server for example. Which is usually the bottleneck? The number of connected sockets? The RAM used by the process? CPU?
>
> Thanks,
>
> Yann
> Le 2 avr. 2013 08:46, "Scott Comer" <we...@mac.com> a écrit :
>> The hardware for those tests that were run years ago is now meaningless. Pentium 4. I've run those tests last year with very different results. The only thing interesting there is the relative performance compared to other tech, which hasn't changed a lot.
>>
>> What matters and nobody else can answer this for you, ever, is how your app performs in your environment.
>>
>> By all means, run the tests. Modify them to more closely model your app and match your needs. As previously stated, see examples/Perf.
>>
>> Scott out
>>
>> Sent from my iPhone
>>
>> On Apr 2, 2013, at 8:37 AM, Siva Kumar <si...@gmail.com> wrote:
>>
>>> Thanks for the reply Martin..
>>>
>>> Scott ,
>>> Please share if you any information on this,
>>>
>>> Regards,
>>> Siva
>>>
>>>
>>> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de> wrote:
>>>> Hi Siva,
>>>>
>>>> > 1. Could please clarify the hardware details,OS details about this test?
>>>> the performance measurements shown in the documentation have been done before I joined the team, so unfortunately I cannot tell you some more details on the hardware or OS used.
>>>> @Scott: Maybe you can help out?
>>>>
>>>> > 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>>>> The performance test itself is available in the example/perf directory, sou you could run them by your own on your specific hardware/OS.
>>>> The IDL description [1] shows the interface (e.g. the "add" function mentioned in the documentation), the implementation is available in [2]. For "add" there are reports for both synchronous and asynchronous calls.
>>>>
>>>>
>>>> Best regards,
>>>> Martin
>>>>
>>>> [1] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>>>> [2] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>>>
>>>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
>>>> Sent: Monday, April 01, 2013 1:01 PM
>>>> To: user@etch.apache.org
>>>> Subject: Etch Performance Details
>>>>
>>>>
>>>>
>>>> Hi All,
>>>>
>>>> I have seen the performance details of Etch in the documentation. I could see the following numbers in the documentation.
>>>>
>>>> System
>>>> Calls/Second
>>>> Etch Java
>>>> 13.3k
>>>> Thrift Java
>>>> 13.5k
>>>> Java RMI
>>>> 7.4k
>>>> CORBA Java
>>>> 2.4k
>>>> JAX-WS
>>>> 943
>>>> Apache Axis2
>>>> 206
>>>>
>>>> 1. Could please clarify the hardware details,OS details about this test?
>>>> 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>>>>
>>>> I am planing to use Etch to my project. Please suggest how to validate the performance.We need to validate the performance before deciding anything on the next step..Please reply as early as possible.
>>>>
>>>> Regards,
>>>> Siva
>>>>
>>>
Re: Etch Performance Details
Posted by Scott Comer <we...@mac.com>.
Obtw, 30k reached with java. Never got very far with .net. Too flakey and immature at the time. Structural problems with performance. Ymmv.
Sent from my iPhone
On Apr 2, 2013, at 10:11 AM, Yann Nicolas <ya...@gmail.com> wrote:
> Hi,
>
> Somebody have measured the maximum concurrent connections we can have (clients connected simultaneously to the server? In a Windows Server 2008 with a C# server for example. Which is usually the bottleneck? The number of connected sockets? The RAM used by the process? CPU?
>
> Thanks,
>
> Yann
> Le 2 avr. 2013 08:46, "Scott Comer" <we...@mac.com> a écrit :
>> The hardware for those tests that were run years ago is now meaningless. Pentium 4. I've run those tests last year with very different results. The only thing interesting there is the relative performance compared to other tech, which hasn't changed a lot.
>>
>> What matters and nobody else can answer this for you, ever, is how your app performs in your environment.
>>
>> By all means, run the tests. Modify them to more closely model your app and match your needs. As previously stated, see examples/Perf.
>>
>> Scott out
>>
>> Sent from my iPhone
>>
>> On Apr 2, 2013, at 8:37 AM, Siva Kumar <si...@gmail.com> wrote:
>>
>>> Thanks for the reply Martin..
>>>
>>> Scott ,
>>> Please share if you any information on this,
>>>
>>> Regards,
>>> Siva
>>>
>>>
>>> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de> wrote:
>>>> Hi Siva,
>>>>
>>>> > 1. Could please clarify the hardware details,OS details about this test?
>>>> the performance measurements shown in the documentation have been done before I joined the team, so unfortunately I cannot tell you some more details on the hardware or OS used.
>>>> @Scott: Maybe you can help out?
>>>>
>>>> > 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>>>> The performance test itself is available in the example/perf directory, sou you could run them by your own on your specific hardware/OS.
>>>> The IDL description [1] shows the interface (e.g. the "add" function mentioned in the documentation), the implementation is available in [2]. For "add" there are reports for both synchronous and asynchronous calls.
>>>>
>>>>
>>>> Best regards,
>>>> Martin
>>>>
>>>> [1] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>>>> [2] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>>>
>>>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
>>>> Sent: Monday, April 01, 2013 1:01 PM
>>>> To: user@etch.apache.org
>>>> Subject: Etch Performance Details
>>>>
>>>>
>>>>
>>>> Hi All,
>>>>
>>>> I have seen the performance details of Etch in the documentation. I could see the following numbers in the documentation.
>>>>
>>>> System
>>>> Calls/Second
>>>> Etch Java
>>>> 13.3k
>>>> Thrift Java
>>>> 13.5k
>>>> Java RMI
>>>> 7.4k
>>>> CORBA Java
>>>> 2.4k
>>>> JAX-WS
>>>> 943
>>>> Apache Axis2
>>>> 206
>>>>
>>>> 1. Could please clarify the hardware details,OS details about this test?
>>>> 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>>>>
>>>> I am planing to use Etch to my project. Please suggest how to validate the performance.We need to validate the performance before deciding anything on the next step..Please reply as early as possible.
>>>>
>>>> Regards,
>>>> Siva
>>>>
>>>
Re: Etch Performance Details
Posted by Yann Nicolas <ya...@gmail.com>.
Hi,
Somebody have measured the maximum concurrent connections we can have
(clients connected simultaneously to the server? In a Windows Server 2008
with a C# server for example. Which is usually the bottleneck? The number
of connected sockets? The RAM used by the process? CPU?
Thanks,
Yann
Le 2 avr. 2013 08:46, "Scott Comer" <we...@mac.com> a écrit :
> The hardware for those tests that were run years ago is now meaningless.
> Pentium 4. I've run those tests last year with very different results. The
> only thing interesting there is the relative performance compared to other
> tech, which hasn't changed a lot.
>
> What matters and nobody else can answer this for you, ever, is how your
> app performs in your environment.
>
> By all means, run the tests. Modify them to more closely model your app
> and match your needs. As previously stated, see examples/Perf.
>
> Scott out
>
> Sent from my iPhone
>
> On Apr 2, 2013, at 8:37 AM, Siva Kumar <si...@gmail.com> wrote:
>
> Thanks for the reply Martin..
>
> Scott ,
> Please share if you any information on this,
>
> Regards,
> Siva
>
>
> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de>wrote:
>
>> Hi Siva,
>>
>> > 1. Could please clarify the hardware details,OS details about this test?
>> the performance measurements shown in the documentation have been done
>> before I joined the team, so unfortunately I cannot tell you some more
>> details on the hardware or OS used.
>> @Scott: Maybe you can help out?
>>
>> > 2. Is this asynchronous/ Non Blocking method call or synchronous
>> /Blocking method call?
>> The performance test itself is available in the example/perf directory,
>> sou you could run them by your own on your specific hardware/OS.
>> The IDL description [1] shows the interface (e.g. the "add" function
>> mentioned in the documentation), the implementation is available in [2].
>> For "add" there are reports for both synchronous and asynchronous calls.
>>
>>
>> Best regards,
>> Martin
>>
>> [1]
>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>> [2]
>> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>
>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
>> Sent: Monday, April 01, 2013 1:01 PM
>> To: user@etch.apache.org
>> Subject: Etch Performance Details
>>
>>
>>
>> Hi All,
>>
>> I have seen the performance details of Etch in the documentation. I
>> could see the following numbers in the documentation.
>>
>> System
>> Calls/Second
>> Etch Java
>> 13.3k
>> Thrift Java
>> 13.5k
>> Java RMI
>> 7.4k
>> CORBA Java
>> 2.4k
>> JAX-WS
>> 943
>> Apache Axis2
>> 206
>>
>> 1. Could please clarify the hardware details,OS details about this test?
>> 2. Is this asynchronous/ Non Blocking method call or
>> synchronous /Blocking method call?
>>
>> I am planing to use Etch to my project. Please suggest how to validate
>> the performance.We need to validate the performance before deciding
>> anything on the next step..Please reply as early as possible.
>>
>> Regards,
>> Siva
>>
>>
>
Re: Etch Performance Details
Posted by Scott Comer <we...@mac.com>.
The hardware for those tests that were run years ago is now meaningless. Pentium 4. I've run those tests last year with very different results. The only thing interesting there is the relative performance compared to other tech, which hasn't changed a lot.
What matters and nobody else can answer this for you, ever, is how your app performs in your environment.
By all means, run the tests. Modify them to more closely model your app and match your needs. As previously stated, see examples/Perf.
Scott out
Sent from my iPhone
On Apr 2, 2013, at 8:37 AM, Siva Kumar <si...@gmail.com> wrote:
> Thanks for the reply Martin..
>
> Scott ,
> Please share if you any information on this,
>
> Regards,
> Siva
>
>
> On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de> wrote:
>> Hi Siva,
>>
>> > 1. Could please clarify the hardware details,OS details about this test?
>> the performance measurements shown in the documentation have been done before I joined the team, so unfortunately I cannot tell you some more details on the hardware or OS used.
>> @Scott: Maybe you can help out?
>>
>> > 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>> The performance test itself is available in the example/perf directory, sou you could run them by your own on your specific hardware/OS.
>> The IDL description [1] shows the interface (e.g. the "add" function mentioned in the documentation), the implementation is available in [2]. For "add" there are reports for both synchronous and asynchronous calls.
>>
>>
>> Best regards,
>> Martin
>>
>> [1] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
>> [2] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>>
>> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
>> Sent: Monday, April 01, 2013 1:01 PM
>> To: user@etch.apache.org
>> Subject: Etch Performance Details
>>
>>
>>
>> Hi All,
>>
>> I have seen the performance details of Etch in the documentation. I could see the following numbers in the documentation.
>>
>> System
>> Calls/Second
>> Etch Java
>> 13.3k
>> Thrift Java
>> 13.5k
>> Java RMI
>> 7.4k
>> CORBA Java
>> 2.4k
>> JAX-WS
>> 943
>> Apache Axis2
>> 206
>>
>> 1. Could please clarify the hardware details,OS details about this test?
>> 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
>>
>> I am planing to use Etch to my project. Please suggest how to validate the performance.We need to validate the performance before deciding anything on the next step..Please reply as early as possible.
>>
>> Regards,
>> Siva
>
Re: Etch Performance Details
Posted by Siva Kumar <si...@gmail.com>.
Thanks for the reply Martin..
Scott ,
Please share if you any information on this,
Regards,
Siva
On Tue, Apr 2, 2013 at 1:15 PM, Martin Veith <Ma...@bmw-carit.de>wrote:
> Hi Siva,
>
> > 1. Could please clarify the hardware details,OS details about this test?
> the performance measurements shown in the documentation have been done
> before I joined the team, so unfortunately I cannot tell you some more
> details on the hardware or OS used.
> @Scott: Maybe you can help out?
>
> > 2. Is this asynchronous/ Non Blocking method call or synchronous
> /Blocking method call?
> The performance test itself is available in the example/perf directory,
> sou you could run them by your own on your specific hardware/OS.
> The IDL description [1] shows the interface (e.g. the "add" function
> mentioned in the documentation), the implementation is available in [2].
> For "add" there are reports for both synchronous and asynchronous calls.
>
>
> Best regards,
> Martin
>
> [1]
> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
> [2]
> http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
>
> From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
> Sent: Monday, April 01, 2013 1:01 PM
> To: user@etch.apache.org
> Subject: Etch Performance Details
>
>
>
> Hi All,
>
> I have seen the performance details of Etch in the documentation. I
> could see the following numbers in the documentation.
>
> System
> Calls/Second
> Etch Java
> 13.3k
> Thrift Java
> 13.5k
> Java RMI
> 7.4k
> CORBA Java
> 2.4k
> JAX-WS
> 943
> Apache Axis2
> 206
>
> 1. Could please clarify the hardware details,OS details about this test?
> 2. Is this asynchronous/ Non Blocking method call or
> synchronous /Blocking method call?
>
> I am planing to use Etch to my project. Please suggest how to validate the
> performance.We need to validate the performance before deciding anything on
> the next step..Please reply as early as possible.
>
> Regards,
> Siva
>
>
RE: Etch Performance Details
Posted by Martin Veith <Ma...@bmw-carit.de>.
Hi Siva,
> 1. Could please clarify the hardware details,OS details about this test?
the performance measurements shown in the documentation have been done before I joined the team, so unfortunately I cannot tell you some more details on the hardware or OS used.
@Scott: Maybe you can help out?
> 2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
The performance test itself is available in the example/perf directory, sou you could run them by your own on your specific hardware/OS.
The IDL description [1] shows the interface (e.g. the "add" function mentioned in the documentation), the implementation is available in [2]. For "add" there are reports for both synchronous and asynchronous calls.
Best regards,
Martin
[1] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/etch/Perf.etch
[2] http://svn.apache.org/repos/asf/etch/trunk/examples/perf/src/main/java/org/apache/etch/examples/perf/MainPerfClient.java
From: Siva Kumar [mailto:sivakumarbe24@gmail.com]
Sent: Monday, April 01, 2013 1:01 PM
To: user@etch.apache.org
Subject: Etch Performance Details
Hi All,
I have seen the performance details of Etch in the documentation. I could see the following numbers in the documentation.
System
Calls/Second
Etch Java
13.3k
Thrift Java
13.5k
Java RMI
7.4k
CORBA Java
2.4k
JAX-WS
943
Apache Axis2
206
1. Could please clarify the hardware details,OS details about this test?
2. Is this asynchronous/ Non Blocking method call or synchronous /Blocking method call?
I am planing to use Etch to my project. Please suggest how to validate the performance.We need to validate the performance before deciding anything on the next step..Please reply as early as possible.
Regards,
Siva