|
>From a Toolbox performance perspective, data queues is about as lightweight as it gets. All work is eventually done by calling an iSeries Data Queue API. In client/server mode the client sends data to the server. The server pulls data off the connection and affects the data queue by calling an iSeries API. In the data queue case little additional processing in needed. If the Java program is running on the same server that holds the queue, there is a native optimization. With the optimization the Toolbox skips the server and directly calling the API (by calling C code via JNI). As with all Toolbox objects, reusing the connection is key when in client/server mode. In client/server mode the best model is connecting up front and reusing the connection until ready to end the application. There was a performance problem with older versions of AS/400. If peek or read with wait was being used (read the queue, if nothing there wait for xxx seconds before returning empty in case something shows up), and an entry was not available, control would not be returned to the client for at least two seconds. That is, even if an entry showed up .1 seconds after the wait, control would still not be returned for two seconds. This was a 'feature' of the server. It took them that long to set up the objects necessary to wait for something to show up. We finally convinced them this was a bug not a feature and they fixed it with a PTF. More info (including PTF numbers) is in the faq section of the Toolbox web site (http://www-1.ibm.com/servers/eserver/iseries/toolbox/faq.htm#faqX). Summary -- data queues is the fastest IPC mechanism available via Toolbox classes. David Wall AS/400 Toolbox for Java "Richards, Etienne (CAP, PTL)" To: "'java400-l@midrange.com'" <java400-l@midrange.com> <Etienne.Richards@ cc: penske.com> Subject: RE: keep a java program open Sent by: java400-l-admin@mi drange.com 01/22/2002 02:51 PM Please respond to java400-l David, What other options do you have for inter process communication on the AS/400? In ILE C you can use shared memory, is that available in RPG? Do you think that shared memory might be faster, all that you then send through the data queues are enough information to identify the shared memory that must be accessed by the receiving application. Can the AS/400 Toolbox people please comment on the speed of accessing data queues from java using the toolbox classes. Etienne -----Original Message----- From: David Morris [mailto:David.Morris@plumcreek.com] Sent: Tuesday, January 22, 2002 2:54 PM To: java400-l@midrange.com Subject: RE: keep a java program open Etienne, Since you mention scalability, I assume you want speed. Data queues in the past with Java were terribly slow. I know there was a fix made that eliminated an unnecessary delay. Data queues are also inefficient with varying length data. I hit that delay problem in an application that passed quite a bit of data. I converted that code to use a user space to pass the bulk of data and a single message on a data queue to trigger an action and it was many times faster. You are right that data queues allow you to scale your application. I built a server framework that uses a configurable number of server jobs and it has worked out very well. It was written in RPGIV, but I am sure that the performance benefit is the same. We use it mostly to farm out FTP jobs, but it also comes in handy for processors we use to sync-up our data warehouse files. David Morris >>> Etienne.Richards@penske.com 01/22/02 12:16PM >>> I have looked at the CPU usage of the java application that is waiting for a message on the queue and it looked good (It seemed like it blocked while waiting, it was not doing polling). Data queues also seemed like the fasted way to get messages to move between different applications that run on the same AS/400 machine. I also feel that this type of solution would scale very well. You can have multiple java programs or multiple java threads read from the same queue and only one of them will get a message, that way you can pre-start more jobs or threads to do the work. For me it felt as if the communication between to two jobs (CL and Java process) was almost instantaneous. It also allowed me to make use of connection pooling for jdbc and AS400 connections which reduced the time to complete the request even more. Are there any other ways to move data from java back to a calling RPG / CL program? I know you can use PCML to call a RPG program from java and you can have input and output parameters but what can you do if you want to call a java program from RPG while wanting to have input and output parameters? You could probably use data areas, MQ Series, Data queues but they all seem to fall into the same type of solution. I hope I answered your question, I know I started rambling a bit... Etienne _______________________________________________ This is the Java Programming on and around the iSeries / AS400 (JAVA400-L) mailing list To post a message email: JAVA400-L@midrange.com To subscribe, unsubscribe, or change list options, visit: http://lists.midrange.com/cgi-bin/listinfo/java400-l or email: JAVA400-L-request@midrange.com Before posting, please take a moment to review the archives at http://archive.midrange.com/java400-l. _______________________________________________ This is the Java Programming on and around the iSeries / AS400 (JAVA400-L) mailing list To post a message email: JAVA400-L@midrange.com To subscribe, unsubscribe, or change list options, visit: http://lists.midrange.com/cgi-bin/listinfo/java400-l or email: JAVA400-L-request@midrange.com Before posting, please take a moment to review the archives at http://archive.midrange.com/java400-l.
As an Amazon Associate we earn from qualifying purchases.
This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].
Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.