[jacorb-developer] Delay in serial communication

[Nick Cross]

Hi,

Reply inline...

On 27/07/12 15:28, Kujtim Hyseni wrote:
>
> Hi Rob,
>
> many thanks for your effort. I will implement your suggestions from
> this and previous emails such:
>
> # Use DNS names in IORs jacorb.dns.enable=off
>
> jacorb.log.default.verbosity=0
>
> What do these mean?

See the JacORB ProgrammingGuide

> By the way how do you measure the time you are mentioned you are
> getting it from ORB by applying jacorb.log.default.verbosity=0 or you
> use some code such timeAfterExec minus timeBeforeExec - can you send
> the code?

Note that any performance measurements should ideally take account of 
'warm up time' and pre-establising connections. Ideally perform them on 
isolated boxes with a dedicated switch.
Also, it very much depends upon what you are trying to measure - for 
instance here are some considerations:
Size of the payload - while there is already an inherent size of payload 
in a TCP message, the CORBA IIOP protocol will add to this.
Time - the time taken to marshal and demarshal your message, plus the 
time taken to send and receive it (simplifying the process ;-) )
Configuration of the ORB e.g. enabling NIO, reducing logging etc.

Regards

Nick


> Thanks again, Kujtim
>
>> Date: Fri, 27 Jul 2012 08:52:26 -0500 From: rrr6399 at futuretek.com
>> To: jacorb-developer at lists.spline.inf.fu-berlin.de Subject: Re:
>> [jacorb-developer] Delay in serial communication
>>
>> FYI:
>>
>> I added some timing loops to retrieving the balance from an account
>> to the Client.java of this example:
>>
>> JacORB/demo/bank/concurrency
>>
>> and got about 2-3 milliseconds on average for a 100 requests on my
>> 5 year old Linux laptop. (I got 1-2 milliseconds with
>> jacorb.log.default.verbosity=0)
>>
>> Rob
>>
>> On 07/27/2012 07:07 AM, Kujtim Hyseni wrote:
>>> I have tested my "methodology" with Web Services (Java SE based),
>>> it worked fine. To illustrate, response times are for 10 requests
>>> (in seconds): 0.156, 0.141, 0.156, 0.141, 0.156, 0.141, 0.140,
>>> 0.157, 0.140 and 0.141.
>>>
>>> So, definitely, the problem is at JacORB. I will upload the code
>>> at bugzilla and hope to get answer from someone soon!
>>>
>>> Kujtim
>>>
>>>> Date: Thu, 26 Jul 2012 08:37:16 -0500 From:
>>>> rrr6399 at futuretek.com To:
>>>> jacorb-developer at lists.spline.inf.fu-berlin.de Subject: Re:
>>>> [jacorb-developer] Delay in serial communication
>>>>
>>>> Have you tried just running one of the JacORB examples to see
>>>> if you get the same behavior such as:
>>>> JacORB/demo/bank/concurrency
>>>>
>>>> On 07/26/2012 08:04 AM, Kujtim Hyseni wrote:
>>>>> Instead of Thread.sleep you may use multiple for/while
>>>>> structures to insert delay. The aim is to illustrate how it
>>>>> affects the system-Windows XP. It blocks everything: task
>>>>> manager, other applications to run etc. It is just to
>>>>> illustrate the effect of jacorb when there are operations to
>>>>> be executed that are time consuming ... Kujtim
>>>> _______________________________________________
>>>> jacorb-developer maillist  -
>>>> jacorb-developer at lists.spline.inf.fu-berlin.de
>>>> https://lists.spline.inf.fu-berlin.de/mailman/listinfo/jacorb-developer
>>>
>>>>
>>> _______________________________________________ jacorb-developer
>>> maillist  -  jacorb-developer at lists.spline.inf.fu-berlin.de
>>> https://lists.spline.inf.fu-berlin.de/mailman/listinfo/jacorb-developer
>>>
>>
>>
>>>
_______________________________________________
>> jacorb-developer maillist  -
>> jacorb-developer at lists.spline.inf.fu-berlin.de
>> https://lists.spline.inf.fu-berlin.de/mailman/listinfo/jacorb-developer
>
>>
> _______________________________________________ jacorb-developer
> maillist  -  jacorb-developer at lists.spline.inf.fu-berlin.de
> https://lists.spline.inf.fu-berlin.de/mailman/listinfo/jacorb-developer
>