Is it possible recording batch jobs using with the load runner tool?

[Narasimha Rao]

Hi Everyone,

This is Rao and i am working as performance tester.

My project is retail domain (i.e. Ecommerce application), in my
project 5 interfaces are there and volumes of data (transactions)
will be move through a middleware (Informatica/Jcaps) on a daily basis
from one interface to another interface through a batch job.

I am facing few problems in my project those are in below.

1.Is it possible to record the batch jobs through a load runner tool?
If possible recording what is the process of recording the scripts?
And what protocol will support for batch jobs recording?

2.If not possible to record the batch jobs using the load runner tool
what is the alternative solution to do performance testing?


I am seeking your help to resolve the above mentioned problems please
help me.

Please let me know if you need any additional information.

Thanks,
Narasimha

[John Crunk]

You want to record batch jobs? My God man! They are batch jobs! Think about
it!!!!

Hi Everyone,

Thanks,
Narasimha

--
You received this message because you are subscribed to the Google
"LoadRunner" group.
To post to this group, send email to LR-Loa...@googlegroups.com To
unsubscribe from this group, send email to
LR-LoadRunne...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/LR-LoadRunner?hl=en

[James Pulley]

Logically your request does not make sense. A batch job would be something
submitted in bulk at a time point or into the backend of a system. It may
be kicked off by a command line operation of execution of a batch script,
but then these same shell commands can be executed by a LoadRunner script.
Precisely how is dependent upon the architecture of the hoist where the
batch is kicked off.

It's time to take a step back and work on cultivating your architectural
eye, a vision which is key to the success of performance testers and
engineers. The smaller your performance test team the more critical it
becomes for your skills to be very precisely honed to be able to tackle
issues such as this one. You might consider speaking with your management
about training, or at the minimum to bring in a mentor to work with you for
the next six months on projects side-by-side.

Otherwise your management is just setting you up to fail and hosing the
shareholders by not obtaining good return on investment for their
performance testing efforts.

James Pulley, http://www.loadrunnerbythehour.com/PricingMatrix

-----Original Message-----
From: lr-loa...@googlegroups.com [mailto:lr-loa...@googlegroups.com]
On Behalf Of Narasimha Rao
Sent: Saturday, April 02, 2011 9:29 PM
To: LoadRunner
Subject: Is it possible recording batch jobs using with the load runner
tool?

Hi Everyone,

Thanks,
Narasimha

--

[pparasher]

Hi Narsimha

Can you please define these batches in greater detail?
What do they do?
When are they Executed?
How are they executed?
Are these batches triggered to run periodically (hourly, nightly,
weekly,etc), or are the constantly running as real-time data feeds?


Depending on the above points, If these are true batch jobs, why not
just write a shell script to automate it. Your Developers might
already have something written in this case. So LR is not required in
this case.

However, if these batches are real-time processes that are executed by
your online traffic, are nt you already testing it as an inherent
function of your application as you run your online users?

Otherwise, you can automate these feeds as part of component level
testing, by understanding the protocols involved and message formats.
(ie- xml over soap, mq,etc)
In this case, can you please provide that by checking with your
Architecture or Dev teams?

Regards,
Puneet

On Apr 2, 10:33 pm, "John Crunk" <jcr...@comcast.net> wrote:
> You want to record batch jobs? My God man! They are batch jobs! Think about

> it!!!!-----Original Message-----

[Dennis]

Hi ,

I have to do performance testing of batch jobs..I would like to know
if i have to measure the performance of batch job then how can i
achieve it.
and what should be the parameters that we should monitor apart from
response time and server side parameters.

I have gone through the below thread.but could not find any mention of
the how to measure the response time while running the batch job.

Kindly let me know..

---------- Forwarded message ----------
From: Narasimha Rao <mnr...@gmail.com>
Date: Apr 2, 6:28 pm
Subject: Is it possible recording batch jobs using with the load
runner tool?

To: LoadRunner

Hi Everyone,

This is Rao and i am working as performance tester.

My project is retail domain (i.e. Ecommerce application), in my
project  5 interfaces are there and volumes of data (transactions)
will be move through a middleware (Informatica/Jcaps) on a daily basis
from one interface to another interface through abatchjob.

I am facing few problems in my project those are in below.

1.Is it possible to record thebatchjobsthrough a load runner tool?

If possible recording what is the process of recording the scripts?
And what protocol will support forbatchjobsrecording?

2.If not possible to record thebatchjobsusing the load runner tool

[John Crunk]

It is possible, but insane to performance test batch jobs. If you find a
problem, how will you improve the performance? Hence the term batch job!!!

-----Original Message-----
From: lr-loa...@googlegroups.com [mailto:lr-loa...@googlegroups.com]

Hi ,

Kindly let me know..

Hi Everyone,

Thanks,
Narasimha

--

[Sagar Aggarwal]

but if we find some problem for e.g. it takes too long to complete batch job processing then we can improve the hardware.I think that's the only thing that we can do.Somebody has told me that we should use stubs to measure response times at various nodes as there is specific tool is needed as these are batch jobs.what is your call on that, folks.

[John Crunk]

How would running the batch job in multiple sessions help find problems?

Thanks,

Dr. John Crunk

Sent from my iPhone

[Sagar Aggarwal]

But cant we check the time it takes to complete batch cycle..I was thinking if batch cycle is taking too much time then we can ask for up-gradation in hardware.

anyways if we are not monitoring batch job processes for performance testing. then i have one query which is mentioned below.

we have to test one portal application and there are many batch jobs running in background .some are daily,some are monthly.

As far as i know we should not ignore these batch jobs, as performance of that portal will be different with or without batch jobs running in background.

So should we run the same load test with or with out batch job process running in background to get a clear picture.

[oc00...@verizon.net]

Sagar, 

 

 Of course you can performance test batch jobs and improve runtimes without necessarily throwing additional hardware at the problem. There are tools, such as Strobe, specifically designed for identifying batch related bottlenecks. However, your batch logs should indicate the runtime for each step. Find the offending step and examine the application associated with it. Many batch jobs can be improved by tuning the SQL or stored procedures associated with the application. Removing contention with other batch jobs running concurrently with yours may also be an option. In addition, there are many other options that may be available to you, depending on how your batch environment is setup and run. Resolving batch performance issues can involve techniques and methods different from those for tuning online user processing.

 

Alan

LR-LoadRunner+u...@googlegroups.com

[Jeff Leonard]

No, but here is how you can leverage your knowledge and expertise.

1. Assess how the file is created and create realistic seed data using VuGen to drop.
2. Fully understand & monitor the architecture once these batch files are dropped. Sometimes there may be a web interface to use. It's a menial loadrunner script and probably the only one run.
3. Most importantly set SLA that the batch cycle needs to adhere to. Then it should be easy to have defects that drive the tuning of the JCAPS architecture.
4. Lastly, if the batch cycle has all night to complete. Aggressively drive down the time it takes to complete. Make them strive to finish as fast as they can. Often batches wait on dependent batches to finish. Driving down your batch time in turn drives down the whole nightly cycle.

That being said, batch does not really leverage good use of the tool but it can be used to solution and your performance engineering knowledge can help optimize this God awful process that we all of hate. We have driven significant change in this area while kicking and screaming the whole way. ;)