Re: PDF Performance

[Peter Connell <Peter.Connell@xxxxxxxxxxxxxxxxx>]

Nathan,
I have mentioned in some previous thread that we have been happily running PD4ML to generate a PDF version of our web page on request by pressing a button.
This has been working well for the past year now and we never have to worry about it. A 10 page PDF response within 5 seconds.

I have not downloaded a newer release (in fact I got an automatic email update from PD4ML yesterday about the latest and greatest).

But I was concerned about performance with regard to loading the JVM.
I was able to determine the appropriate class to call with the intention of calling a class directly from RPG to benefit from the 1 JVM per job feature.
In fact I found a bit of java code on the web somewhere that showed how to create a little wrapper that called the correct PD4ML class in the expected manner.
So I compiled this wrapper as a little class of my own which I now call from RPG.
To avoid the JVM startup for the job I use a server job and a data queue (I know you'll like that)

Running the class in a server job was also essential since, to implement the headless java requirement, IBM apparently launches a PASE environment within the job just to achieve this.
Unfortunately this conflicts with any existing apps which may already run some task that uses PASE because they fail when they attempt to start the PASE environment.

Peter



-----Original Message-----
From: web400-bounces@xxxxxxxxxxxx [mailto:web400-bounces@xxxxxxxxxxxx] On Behalf Of Nathan Andelin
Sent: Wednesday, 12 May 2010 9:07 a.m.
To: Web Enabling the AS400 / iSeries
Subject: [WEB400] PDF Performance

One thing that motivated me to download pd4ml was a performance benchmark, published on their web site, which indicated much higher throughput than that of my testing with Aaron Bartell's HTML2PDF4i utility.  My original thought was that a performance advantage alone might justify the additional cost of a pd4ml license.  However, both utilities appear to offer identical performance in generating an identical 150 page PDF document on my server.

That may not be saying much.  It takes 2 minutes to generate that PDF on my server.  To be honest, that kind of performance is a bit repulsive to me.  I've done a number of performance tests, and I think these Java jars may be a case where my 1 Gig of RAM, is insufficient.

Performance may be a moot point, however.  PDF reports are a requirement!

Any thoughts about performance?

-Nathan.