If you and/or your third party software use a log facility (I think most
software do) such as log4j you can turn on debug mode and inspect the log’s timestamp and see where the time goes Easier said than done, especially for big systems. Also, as a rule, requires some instrumentation to be included in the application being tested. Otherwise, only servlet/enterprise bean-level information would be available - personally, I find that insufficient. In any case, I certainly did not mean that type of 3rd party software. Lo ________________________________________________________ Misys Banking Systems Key West / Windsor Road / Slough / SL1 2DW T +44 (0)1753 708 053 M +44 (0)7903 181 030 raikovl1@xxxxxxxxx www.misys.com ________________________________________________________________ This email and any attachments have been scanned for known viruses using multiple scanners. We believe that this email and any attachments are virus free, however the recipient must take full responsibility for virus checking. This email message is intended for the named recipient only. It may be privileged and/or confidential. If you are not the intended named recipient of this email then you should not copy it or use it for any purpose, nor disclose its contents to any other person. You should contact Misys Banking Systems so that we can take appropriate action at no cost to yourself. www.misys.com <http://www.misys.com/> -----Original Message----- From: Bruce Jin [mailto:brucej@xxxxxxxxxxxxxxxxxxxx] Sent: 17 October 2006 14:39 To: Java Programming on and around the iSeries / AS400 Subject: Re: Optimizing Native Java If you and/or your third party software use a log facility (I think most software do) such as log4j you can turn on debug mode and inspect the log’s timestamp and see where the time goes regards. Raikov, Leonid wrote:
The most important thing is to understand where the time goes. Situations when there are no bottlenecks and every class contributes to poor performance in equal measure are pretty rare. What you need is to a) understand what threads are primarily involved in productive work and then b) draw a chart describing what each of these thereads is busy doing (e.g. 50% of the time spent waiting for JDBC queries and 50% of the time doing
processing). You can hardly achieve this without Java performance analysis tools for System i. IBM Job Watcher is what immediately springs to mind,
JW data is not easy to analyse, to say the least. There are other tools, some of them pretty good, but none of them in the Open Source domain, as
as I know. As for the magic button, there is none, I'm afraid. I would be suprised if memory alone could make you happy. That is, unless you're running your JVM in a 50MB pool. Lo -----Original Message----- From: albartell To: 'Java Programming on and around the iSeries / AS400' Sent: 10/16/06 4:12 PM Subject: Optimizing Native Java Hi All, I am just completing a project where I wrote a Java proxy of sorts and there were a bunch of third party jar files involved in the mix. To make a long story short we are not satisfied with the performance we are getting out of the Java processes on the iSeries (note the Java is running in it's own job listening to a data queue - only one JVM startup). We have run the CRTJVAPGM against all .class and .jar files involved and are wondering are there any other mechanisms to look at concerning getting more speed on the Java end? Obviously one solution would be to throw more memory at it, but is there a way to section off the memory so this Java process gets sole use of it (similar to how you can do the same for Websphere Application Server)? Any other idea's? Aaron Bartell New Tool! - RPG Chart Engine - visit www.mowyourlawn.com <http://www.mowyourlawn.com/> for more info.
As an Amazon Associate we earn from qualifying purchases.
Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.