Re: How much real is required for 64-bit
Not to mention with a wait stated system, since you cannot 'attach a zillion page packs'. You are limited to 253x4GB page space, about one TB. Having so many page space makes one want to use dynamic PAV volumes, it means 253x2 device devices addresses, and it means maximum 128 base volumes in a logical CU. If we take a 5-way sysplex, we need 5x253x2 device addresses for page datasets. Zaromil -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
>Unfortunately, with a max of 128 gig of real, that would leave >you with the requirement to attach a zillion page packs to the system. >Performance would probably not be too good. Not to mention with a wait stated system, since you cannot 'attach a zillion page packs'. You are limited to 253x4GB page space, about one TB. Regards, Barbara Nitz -- Highspeed-Freiheit. Bei GMX supergünstig, z.B. GMX DSL_Cityflat, DSL-Flatrate für nur 4,99 Euro/Monat* http://www.gmx.net/de/go/dsl -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
Thanks for everyone for your suggestion. Niran Kamaksorn Bangkok, Thailand -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
At several SHAREs I gave a pitch on this conversion. The last time was in Boston to a nearly empty room: I'm done with it. Check the proceedings for Session 2839 SHARE 105. . . . JO.Skip Robinson Southern California Edison Company SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [EMAIL PROTECTED] IBM Mainframe Discussion List wrote on 11/10/2005 11:27:48 PM: > Hi All, > We're going to do 64bit conversion soon (z/OS 1.4) so could you guys please > suggest us the steps/ways for this conversion and what kind of applications > that should be test to assure if this conversion is work ? And anything > should be considered as a problem that we may need to aware of it ? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
In our case, we saw no problem whatsoever in moving from 32-bit OS/390 2.10 to 64-bit z/OS 1.4. We did, however, hit a couple of glitches (in ISV software -- Model 204 and CA-Dispatch) when we finally increased the amount of physical memory allocated to our production LPAR to more than 2GiB. Jon As far as 64-bit bugs with more than 2G real, as Chris pointed out the IBM bugs have been fixed for years. So have the ISV bugs. So unless you are running very old ISV software, I personally wouldn't give it a second thought. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
On Thu, 10 Nov 2005 09:54:26 -0500, Clark, Kevin D, HRC-Alexandria/EDS <[EMAIL PROTECTED]> wrote: >Question: > >I have a few lpars with 192MB of real defined. Would that be enough for >64-bit testing ? > The absolute minimums are documented in the Planning for Installation manual (see cut/paste from section 4.3.1.2 - z/OS 1.4 below). As far as 64-bit bugs with more than 2G real, as Chris pointed out the IBM bugs have been fixed for years. So have the ISV bugs. So unless you are running very old ISV software, I personally wouldn't give it a second thought. As far as Skip's points about in-house programs that may fail, he is on target, but I haven't seen this to be an issue in any of the conversions I was personally involved in - and while consulting I was involved in lots of them at many different shops (perhaps he and others have though). I ran into one 64-bit problem at one client with an IBM tool for monitoring LLA (MFM - Module Fetch Monitor) and Peter Relson fixed that in 2002. Even if you have a surplus of memory and can add enough to a test environment so there is more than 2G real, can you identify and test every home grown program in the shop? Is it worth the trouble even if you could? At this point I would take a leap of faith and just make sure you have a backout plan. As long as you are current on maintenance, I highly doubt you'll be backing out due to a 64-bit issue. If your conversion to 64-bit is coupled with an OS upgrade at the same time, it is much more likely you'll have issues realted to that. Here is the doc from the z/OS 1.4 planning for installation on the minimum processor storage required to IPL: 4.3.1.2 Processor storage Expanded storage is not supported in z/Architecture. Thus, you must configure your LPAR to use only central storage (and not expanded storage). The minimum processor storage required to IPL z/OS or z/OS.e depends on a number of factors, including the number of devices in the I/O configuration, the amount of storage configured, the number of address spaces created, the types of applications supported, and the products and subsystems supported in the IPLed system. Nevertheless, we can make the following statements about the minimum processor storage required to IPL: It does not usually change from OS/390 release to OS/390 release, z/OS release to z/OS release, or z/OS.e release to z/OS.e release. It increases when you make the jump from OS/390 to z/OS or z/OS.e. This is due to the change in architecture (from 31-bit ESA/390 architecture to 64-bit z/Architecture) when you run z/OS on a z800 or z900 server, or z/OS.e on a z800 server. It should be the same for z/OS.e as for z/OS. In IPL studies at IBM, 24 MB were required to IPL the last three OS/390 releases (R8, R9, and R10). Also in the studies, 35 MB were required to IPL the first two z/OS releases (V1R1 and V1R2). The systems in the studies were IPLed to the point of completing JES initialization. The systems had I/O configurations defined with the following devices: Device Quantity 3270 32 3390 1152 3490 16 In additional studies with different I/O devices, 42 MB were required to IPL z/OS V1R3 and z/OS V1R4. The systems in the studies were IPLed to the point of completing JES initialization. The systems had I/O configurations defined with the following devices: Device Quantity 3270 32 3390 3264 3490 16 OSA 32 Cheers, Mark -- Mark Zelden Sr. Software and Systems Architect - z/OS Team Lead Zurich North America and Farmers Insurance Group mailto: [EMAIL PROTECTED] Systems Programming expert at http://Search390.com/ateExperts/ Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
How much real is required for 64-bit
> I have a few lpars with 192MB of real defined. Would that be enough for > 64-bit testing ? The discussiong seemed to get off track about real storage. You can start testing 64 bit virtual with any amount of real storage. It is virtual storage and only what you use, requires backing in real or aux. Start out with simple C programs that you can compile for 64 bit. They will use a little more virtual than the 31 bit version, but simply compiling for 64 bit will not cause them to require significantly more real storage to run. What you don't want to do is create a testcase which allocates 20 gig of 64 bit virtual and then attempts to back all of the pages by touching them. For that to work, you would need at least 20 gig of real plus paging space. Within IBM, we have run tests with a 64 bit JVM and an 8 gig java heap. It works fine, but some testers thought that if one was good, why not start 10 of them. To do that, you simply need more real. Most customers don't run with too much paging. That remains true with 64 bit virtual as well. The capability is there to write a program that allocates 16 exabytes of virtual. Unfortunately, with a max of 128 gig of real, that would leave you with the requirement to attach a zillion page packs to the system. Performance would probably not be too good. Don Ault, 8-295-1750, 845-435-1750 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
One problem we encountered was with CA1 and CA90's. We were at an older version(s) and had to put maintenance on pronto for our production side. When we tested all apps on the TEST lpar, all was okay, but we in fact did not have the 2 gig of memory over there, so the one CA problem did not arise until we went to production. HTH, Claude >>> [EMAIL PROTECTED] 11/11/05 2:27 AM >>> Hi All, We're going to do 64bit conversion soon (z/OS 1.4) so could you guys please suggest us the steps/ways for this conversion and what kind of applications that should be test to assure if this conversion is work ? And anything should be considered as a problem that we may need to aware of it ? Niran Kamaksorn Bangkok, Thailand -IBM Mainframe Discussion List wrote: - >To: IBM-MAIN@BAMA.UA.EDU >From: Barbara Nitz >Sent by: IBM Mainframe Discussion List >Date: 11/11/2005 13:12 >Subject: Re: How much real is required for 64-bit > >Clark, > >back when we did 64bit conversion we started out on an lpar that was >real >small (way below 2GB). We ran into a few problems because the lpar >was so >small that no one else had seen before us. (But that was OS/390 >2.10.) > >Best regards, Barbara Nitz > >-- >10 GB Mailbox, 100 FreeSMS/Monat http://www.gmx.net/de/go/topmail >+++ GMX - die erste Adresse f?r Mail, Message, More +++ > >- >- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN >INFO >Search the archives at http://bama.ua.edu/archives/ibm-main.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html -- The contents of this e-mail (and any attachments) are confidential, may be privileged and may contain copyright material. You may only reproduce or distribute material if you are expressly authorized by us to do so. If you are not the intended recipient, any use, disclosure or copying of this email (and any attachments) is unauthorized. If you have received this e-mail in error, please notify the sender and immediately delete this e-mail and any copies of it from your system. == -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
Skip Robinson wrote: By all means *start* your 64-bit testing on the most disposable LPAR you have regardless of size. However, certain 64-bit problems *will not* show up in an LPAR that has less than 2 gig of memory. Before going to production in a >2 gig LPAR, you must somehow build a large test LPAR if only temporarily. Otherwise you may be in for some nasty surprises. Usually programs, even poorly written use virtual memory, so wouldn't be so afraid about potential problems. (HINT: system does use real memory, but it is tested and patched quite well since 2000). BTW: Let's assume everything works fine on 192MB real memory. Now we get into (let's assume again: SMALL) production, with prod LPAR has less than 2GB assigned. Why to test issues "above 2 GB" ? BTW2: The only memory problems I ever had were related to (virtual) storage below 16MB, doesn't matter is was 31-bit OS/390 or 64-bit z/OS. -- Radoslaw Skorupka Lodz, Poland -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
Hi All, We're going to do 64bit conversion soon (z/OS 1.4) so could you guys please suggest us the steps/ways for this conversion and what kind of applications that should be test to assure if this conversion is work ? And anything should be considered as a problem that we may need to aware of it ? Niran Kamaksorn Bangkok, Thailand -IBM Mainframe Discussion List wrote: - >To: IBM-MAIN@BAMA.UA.EDU >From: Barbara Nitz >Sent by: IBM Mainframe Discussion List >Date: 11/11/2005 13:12 >Subject: Re: How much real is required for 64-bit > >Clark, > >back when we did 64bit conversion we started out on an lpar that was >real >small (way below 2GB). We ran into a few problems because the lpar >was so >small that no one else had seen before us. (But that was OS/390 >2.10.) > >Best regards, Barbara Nitz > >-- >10 GB Mailbox, 100 FreeSMS/Monat http://www.gmx.net/de/go/topmail >+++ GMX - die erste Adresse fr Mail, Message, More +++ > >- >- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN >INFO >Search the archives at http://bama.ua.edu/archives/ibm-main.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
Clark, back when we did 64bit conversion we started out on an lpar that was real small (way below 2GB). We ran into a few problems because the lpar was so small that no one else had seen before us. (But that was OS/390 2.10.) Best regards, Barbara Nitz -- 10 GB Mailbox, 100 FreeSMS/Monat http://www.gmx.net/de/go/topmail +++ GMX - die erste Adresse für Mail, Message, More +++ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
>Maybe, but the RSM issues have been fixed for years now. I would expect that simply applying the necessary maintenance would be sufficient to avoid most of the problems. Much the same would apply with ISV software by now. ... Boy! Aren't you the optimist! I have been burned (recently) by 'FIXED' software. The SYSPROG's #1 rule: “All software sucks!”. -teD Me? A skeptic? I trust you have proof! -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
>As best as I can tell, the 2Gib to 4Gib "twilight zone" is strictly virtual and a z/OS only "twilight zone" ... It is strictly virtual. You can verify it for yourself with an RMF report. Count the online frames, the various buckets and the afq. Left side == Right side. QED -teD Me? A skeptic? I trust you have proof! -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
>I have a few lpars with 192MB of real defined. Would that be enough for 64-bit testing ? ... I wouldn't attempt it without at least some memory above 2GB. Otherwise, how would you know anything was working? -teD Me? A skeptic? I trust you have proof! -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
The nasty surprises are hiding in your own code--or perhaps that super handy utility you found on the CBT tape 'sometime in the 80s or 90s and today could hardly live without. Also hiding in that pesky little backwater application that's been off support for 12 years but the clients just can't find the time/money/motivation to convert over to something more modern. You held your breath on 1/1/2000 and were relieved to find that these charming relics survived splendidly. Now run them on an LPAR that actually makes use central storage above the bar. They may survive again. Or not. . . . JO.Skip Robinson Southern California Edison Company SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [EMAIL PROTECTED] IBM Mainframe Discussion List wrote on 11/10/2005 11:04:55 AM: > > > Before going to production in a >2 gig > > > LPAR, you must somehow build a large test LPAR if only > > > temporarily. Otherwise you may be in for some nasty surprises. > > Maybe, but the RSM issues have been fixed for years now. I would expect > that simply applying the necessary maintenance would be sufficient to > avoid most of the problems. Much the same would apply with ISV software > by now. > > > Isn't there a "twilight zone" between 2 GiBi and 4 GiBi? or does that > > apply only to virtual storage? > > Only to virtual. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
Skip Robinson wrote: By all means *start* your 64-bit testing on the most disposable LPAR you have regardless of size. However, certain 64-bit problems *will not* show up in an LPAR that has less than 2 gig of memory. Before going to production in a >2 gig LPAR, you must somehow build a large test LPAR if only temporarily. Otherwise you may be in for some nasty surprises. If you have z/VM, define your real storage discontiguously. When we first tested our 64-bit support, we defined two 192MB real memory segments for the z/OS guest: 0 through 192M and 2G through 2G+192M for a total of 384MB of real. -- - | Edward E. Jaffe|| | Mgr, Research & Development| [EMAIL PROTECTED]| | Phoenix Software International | Tel: (310) 338-0400 x318 | | 5200 W Century Blvd, Suite 800 | Fax: (310) 338-0801| | Los Angeles, CA 90045 | http://www.phoenixsoftware.com | - -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
> > Before going to production in a >2 gig > > LPAR, you must somehow build a large test LPAR if only > > temporarily. Otherwise you may be in for some nasty surprises. Maybe, but the RSM issues have been fixed for years now. I would expect that simply applying the necessary maintenance would be sufficient to avoid most of the problems. Much the same would apply with ISV software by now. > Isn't there a "twilight zone" between 2 GiBi and 4 GiBi? or does that > apply only to virtual storage? Only to virtual. CC -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
> -Original Message- > From: IBM Mainframe Discussion List > [mailto:[EMAIL PROTECTED] On Behalf Of Paul Gilmartin > Sent: Thursday, November 10, 2005 12:24 PM > To: IBM-MAIN@BAMA.UA.EDU > Subject: Re: How much real is required for 64-bit > > > I'd guess the "twilight zone" is virtual; I'd hope the system design > wouldn't squander 2 GiB of real storage (or are the sockets between > 2GiB and 4GiB simply left empty; no DIMMs installed?) > > Does something, then, have a problem with real addresses with the > sign bit set? What? What software is so concerned with real > addresses other than OS memory management, which should have been > thoroughly tested at other sites by now? > > -- gil > -- As best as I can tell, the 2Gib to 4Gib "twilight zone" is strictly virtual and a z/OS only "twilight zone". The problem is the fact that putting the high order bit of an AMODE(31) address "ON" is an "end of list" indication when using the standard linkage conventions. The "twilight zone" is 0x00 00 00 00 80 00 00 00 to 0x00 00 00 00 FF FF FF FF because of the question of "how do I interpret these addresses?" Are they actually address between 2Gib and 4Gib, or are the below 2Gib addresses with the high bit set for "end of address"? The z/OS designers simplified this by simply x'ing out this area. I guess that is why it is called "above the bar". The "bar" is 2Gib in size. There is nothing in the hardware itself, as I read the zPOPS, which "dislikes" those addresses. -- John McKown Senior Systems Programmer UICI Insurance Center Information Technology This message (including any attachments) contains confidential information intended for a specific individual and purpose, and its' content is protected by law. If you are not the intended recipient, you should delete this message and are hereby notified that any disclosure, copying, or distribution of this transmission, or taking any action based on it, is strictly prohibited. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
In a recent note, Chase, John said: > Date: Thu, 10 Nov 2005 12:11:03 -0600 > > > -Original Message- > > From: IBM Mainframe Discussion List On Behalf Of Skip Robinson > > > > By all means *start* your 64-bit testing on the most > > disposable LPAR you have regardless of size. However, certain > > 64-bit problems *will not* show up in an LPAR that has less > > than 2 gig of memory. Before going to production in a >2 gig > > LPAR, you must somehow build a large test LPAR if only > > temporarily. Otherwise you may be in for some nasty surprises. > > Isn't there a "twilight zone" between 2 GiBi and 4 GiBi? or does that apply > only to virtual storage? > I'd guess the "twilight zone" is virtual; I'd hope the system design wouldn't squander 2 GiB of real storage (or are the sockets between 2GiB and 4GiB simply left empty; no DIMMs installed?) Does something, then, have a problem with real addresses with the sign bit set? What? What software is so concerned with real addresses other than OS memory management, which should have been thoroughly tested at other sites by now? -- gil -- StorageTek INFORMATION made POWERFUL -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
> -Original Message- > From: IBM Mainframe Discussion List On Behalf Of Skip Robinson > > By all means *start* your 64-bit testing on the most > disposable LPAR you have regardless of size. However, certain > 64-bit problems *will not* show up in an LPAR that has less > than 2 gig of memory. Before going to production in a >2 gig > LPAR, you must somehow build a large test LPAR if only > temporarily. Otherwise you may be in for some nasty surprises. Isn't there a "twilight zone" between 2 GiBi and 4 GiBi? or does that apply only to virtual storage? -jc- -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How much real is required for 64-bit
By all means *start* your 64-bit testing on the most disposable LPAR you have regardless of size. However, certain 64-bit problems *will not* show up in an LPAR that has less than 2 gig of memory. Before going to production in a >2 gig LPAR, you must somehow build a large test LPAR if only temporarily. Otherwise you may be in for some nasty surprises. . . . JO.Skip Robinson Southern California Edison Company SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [EMAIL PROTECTED] IBM Mainframe Discussion List wrote on 11/10/2005 06:54:26 AM: > Question: > > I have a few lpars with 192MB of real defined. Would that be enough for > 64-bit testing ? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
How much real is required for 64-bit
Question: I have a few lpars with 192MB of real defined. Would that be enough for 64-bit testing ? Kevin -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
