Ed, 1) I know that using REXX "spawn", it will use an environmental variable, "_BPX_SHAREAS=YES", to decide whether to run in the same address space. The same may be true for "fork" and C++.
2) We are running 2 production WebSphere/DB2 and 1 production Perl/Apache/PostgreSQL applications on zLinux. I'd be willing to share my experiences. Thanks, Hank Calzaretta Moore Wallace (630)799-2436 -----Original Message----- From: Edwin Handschuh [mailto:[EMAIL PROTECTED] Sent: Thursday, February 19, 2004 3:55 PM To: [EMAIL PROTECTED] Subject: REAL WORLD L390 APPLICATIONS? Hi all: Background: My client has developed an application that runs under MVS/USS. It issues a FORK instruction which causes MVS to create a new address space for the new task. Considering the new application is going to be running approximately 5,000 tasks simultaneously, this presents a significant amount of overhead. I'm told a similar application running on an HP-UX environment (or other pure Unix environment I guess) wouldn't have the "new address space" overhead to contend with and the application would perform reasonably well. Yes, I know, this is clearly an application that should be running under CICS, but hey, I didn't design it. I was called in after the fact. At least it isn't in production... yet. Naturally, everyone is ready to trash the mainframe and go with a Unix solution on a Unix platform, but my client would prefer the application reside on the mainframe (too many reasons to explain herein). Enter L390: I suggested we take a look at L390. You can imagine the death stares I received. It went over like a lead balloon. The primary resistance stems from a common belief that "NOBODY's RUNNING LINUX TO DO REAL WORK" on the mainframe. My Questions: 1. Is anybody out there using L390 for "production" application workloads? 2. If so, are you willing to share some of your experiences and provide "real world" examples? Oh, one last thing: There seems to be a belief that L390 is a port of meta code instead of natively compiled code. My guess is that its heavily tweaked code compiled to run on z/Series hardware. In fact, the more that I think about it, the more I know it has to be rather unique considering the I/O subsystem. TIA. Ed Handschuh Enterprise Operating Systems Architect Independent Consultant SoftExcell, Inc. (215) 783-2208 - cell
