Perf Time: Layer Wise Breakdown

On Mar 8, 2010, at 6:44 PM, Rahul Bhatt wrote:
> I have been given a task to calculate time spent in each tier of the
> application. The application is web based j2ee application which
> uses oracle in backend.
>
> I have used Fiddler, Loadrunner and HP Diagnostic to calculate these
> timings.

> Fiddler - total time (t1)
> Loadrunner - Web + App + DB (t2)
> HP Diag - App (t3) + DB (t4)

> Client time (IE rendering + network) = t1 - t2
> Web tier = t2 - t3 - t4
> Application tier = t3
> Database = t4

> I would appreciate if this can be validated by experts in the forum.
> All suggessions are very well welcome.
>
> Thanks
****************************************************************************
-----Original Message-----
From: LoadRunner@yahoogroups.com [mailto:LoadRunner@yahoogroups.com] On Behalf Of James Pulley
Sent: Tuesday, March 09, 2010 10:15 AM
To: LoadRunner@yahoogroups.com
Subject: RE: [LoadRunner] Time spent in each j2ee tier

I have a far simpler proposition if you want to know how much time is spent inside the java tier (assuming that this is on a dedicated host)
(1) Grab a protocol analyzer
(2) Set your filters to include traffic to and from the Java host from the next upstream and downstream architectural components.
(3) begin recording conversations
(4) Reassemble streams into sessions using your protocol analyzers capabilities
(5) Look at timestamps (to the millisecond) from the last frame of the request to the first frame of the response. The difference is the time spent in your Java tier.


And it works in production too.

As to Fiddler? Have you considered the advantage of the Graphical Virtual User technology which will not only collect the response times inside of LoadRunner, but you can then present the data compared to API calls on the same graph, where the difference is time in client (which includes client processing plus rendering)

****************************************************************************
Alternate path. Dump your Java Server logs. Analyze the information with Microsoft LogParser for requests and responses data.