Facing Problem after IBM MQ Installation on Windows NT Version 5.2. 1
BDY.RTF Description: RTF file
Re: WMQ Conversion
Conversion for the GET requested by MQGMO_CONVERT is done at the receiving end. Specifying CONVERT(YES) is done at the sending end of the channel. Regardless of the option selected the MQMD is always converted if necessary. So if you are running on two systems that have the same hardware the the cpu utilisation numbers for either option would be almost identical. For disparate systems then it becomes a matter of where you want the work to take place. For example S390 has a special translate(TR) instruction built into the CISC but it might still pay to have the conversion done on the sending UNIX system. That said there is a strong convention of receiver makes good, i.e. the receiver does the conversion. The primary reason given for this is that where a message has to travel through multiple systems before it reaches its target queue converting it at each step would actually slow the process down. FInally given that S390 MIPS dont equate to HP MIPS dont equate to Intel clock rates if you are seriously considering using CONVERT(YES) on you sender channels I'd suggest you test the round trip response time of a series of request-reply messages on the systems involved. I dont know if there is a performance report on this at the IBM supportpac site but it might be worth having a quick look. Regards Tim A Dawson, John [EMAIL PROTECTED]To: [EMAIL PROTECTED] IGROUP.COM cc: Sent by: MQSeries Subject: WMQ Conversion List [EMAIL PROTECTED] .AC.AT 16/03/2003 03:29 Please respond to MQSeries List Fellow MQ'ers Does anyone have any information regarding the CPU utilization of converting a message using channel conversion vs. converting a message with the get option. Are there any supportpacs that have CPU utilization studies for message conversion? Any real time experience? Thanks, John Dawson Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive
Re: SWOT Analysis with listeners as inetd or runmqlsr AND channels as threads or processes
Inetd for more than a couple of hundred connections is usually more
reliable. Runmqlsr and threads uses less resources. As for your second
question its determined by which type of listener you use. To quote from
the manual.
You can use inetd or the Run Listener (RUNMQLSR) command to define a TCP/IP
connection on a UNIX systerm server, . If you use inetd, a process sis
started for each connection you define. If you use the RUNMQLSR command, a
thread is started for each connection. This method can therefore be more
efficient.
I have seen both working well on small systems, however for systems that
have several thousand client connections we use inetd.
Regards
Tim A
Stephan C. Moen
[EMAIL PROTECTED]To: [EMAIL PROTECTED]
Sent by: MQSeriescc:
List Subject: SWOT Analysis with listeners
as inetd or runmqlsr AND channels as
[EMAIL PROTECTED] threads or processes
N.AC.AT
15/03/2003 16:03
Please respond to
MQSeries List
MQSeries Experts,
I am inquiring from the vast array of knowledge within the MQSeries
community on two simple topics. Please respond to the strengths and
weaknesses of the following.
1) Choice of listener: inetd or runmqlsr process.
2) Choice of channel: start as a thread or process.
Im not looking for book responses, just REAL-LIFE experiences, especially
from a performance, reliability, scalability, and MQSeries Version (5.3,
5.2 and below) perspective. Thank you.
Steve Moen
Ëk¹Ëb¢{ ¢¹¨ ¨º¹X§X¬¶±Ê⦪º/{ax¸¬¶¼g§z¶¥R°k¢uæ¯j)ZnW¶m§ÿðÃ
l¡û\¢`+r¯zm§ÿ!Â'§iüÄz¸±ª+Þ
Strange Problem - Setting up MQCICSBridge
Hi, We encountered this problem recently when setting up the MQCICSBridge component on the cics region. We're using mq 5.2 on os390 and after applying the components on the cics region(v4.1) (ie define the pgms, trans id ckbr, ckti etc, define the loadlib mqm.*), the current application PL1 COBOL(map screen) having PROG755,PROG753 error. Temporary fix by removing the loadlib mqm.* from the cics and it works fined. It has been reported to IBM for futher investigation. DOes anyone out there ever encounter this problem and would like to share the solution on this? Thanks, Joe __ Do you Yahoo!? Yahoo! Web Hosting - establish your business online http://webhosting.yahoo.com Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive
