Re: Old discussion about Windows running on a mainframe ( I brought up)
Platform Solutions, Inc. had this technology and much more. Their technology is now IBM's. Many people accuse IBM of acquiring PSI to stifle competition, but this is a great disservice to the engineers at PSI, particularly former Amdahl engineers, who are now bound to silence by NDAs. http://www.nytimes.com/2009/03/23/technology/companies/23mainframe.html?_r=3pagewanted=2ref=technologyadxnnlx=1237816804-Ls%2021S/ySzdFWiPCcF03qQ IBM may see the virtualization of Windows onto z architecture as strategic; both as servers and workstations and its acquisition of PSI reflects this. From what little Mantissa has disclosed, their approach, IMHO, is inherently flawed. Their SHARE presentation seemed to be little more than an obfuscated discourse on virtualization in general. Even if their approach works, it would have been better, architecturally, to emulate an Itanium for Windows hosting because of the nature of the Itanium’s instruction parallelism. A more sensible approach would be to look at creating a Windows HAL (hardware abstraction layer), or something conceptually similar, that runs on z/Series. Historically, this is how Windows has been made to run on different machine architectures. Of course, cooperative development between IBM and Microsoft would be necessary. Another possibility is to exploit the Infiniband feature of the z/10. This feature is profound in terms of 360-z/series evolution, but has been largely ignored, so far. Infiniband attached external hardware products that expose x86 architecture processors, from Intel or IBM itself (or maybe even Intel Larrabee), is an ideal way to run Windows on IBM mainframes. In the interest of disclosure, I have worked for both IBM and PSI, but the opinions I express here are complete conjecture. Harry harry_w...@hotmail.com Date: Mon, 23 Mar 2009 21:08:01 -0700 From: ps2...@yahoo.com Subject: Old discussion about Windows running on a mainframe ( I brought up) To: IBM-MAIN@bama.ua.edu Will Big Blue mainframes run Windows? Track this topic Print story Post comment z/VMs get Microsoft rumor By Timothy Prickett Morgan • Get more from this author Posted in Servers, 23rd March 2009 22:34 GMT Whitepaper download - Eight CRM essentials An obscure mainframe software company called Mantissa Corporation bragged last summer on the IBM VM listserv - which is dedicated to virtual mainframe environments - that it was creating a product called z/VOS that would allow slices of a Windows operating system to run atop z/VM, the hypervisor-as-operating system for IBM mainframes. The product was due in the first quarter of this year, and the story of its impending release has been making the rounds. According to a report in NetworkWorld, Mantissa's z/VOS, presumably short for Virtual Operating System, is a layer of software for VM that allows desktop and server Windows operating systems to run in emulated mode atop z/VM. Mantissa - which is based in Birmingham, Alabama, and which is a supplier of report distribution and other tools for mainframes - talked about the z/VOS product at the SHARE mainframe user conference in early March in Austin, Texas. But that was not the same thing as a product launch. We've tried to reach the company for several days, but Mantissa has yet to respond. While IBM and the Linux community for mainframes centered around Marist College in New York have worked to get official mainframe ports done for Linux - Red Hat and Novell officially support mainframes, if you can write a big enough check to get support - there is no native Windows port to IBM mainframes as far as I know. So, the real curiosity is how Mantissa is supporting Windows XP or Vista atop z/VM partitions. According to the company's development blog, z/VOS includes a translation engine that converts native x86 code to its System z equivalent. See how easy that was? As it translates equivalent results - not creating equivalent machine code, mind you - the instruction that is created by z/VOS is stored in memory so it can be accessed the next time the operating system function inside Windows running on the mainframe is asked for again. Since Gary Dennis, Mantissa's chief executive officer and founder - and other we've called - are not answering their phones, it is a little hard to take the company seriously. But if it can indeed deliver a layer of abstraction software atop z/VM that lets Windows desktops and servers run on mainframe iron, the company should probably think about getting someone to answer the phones and maybe a salesperson or two to try to take some orders. If the x86 translation overhead is not too high, this could be a very interesting development - and one that Big Blue would seem pretty keen on supporting, not quashing. ® http://www.theregister.co.uk/2009/03/23/mantissa_windows_on_mainframes/ AT least these people
Re: Old discussion about Windows running on a mainframe ( I brought up)
On Tue, Mar 24, 2009 at 7:35 AM, Harry Wahl harry_w...@hotmail.com wrote: Platform Solutions, Inc. had this technology and much more. Their technology is now IBM's. Doesn't PSI does the opposite of what Mantissa claims to be working on? PSI runs z architecture on x86 hardware whereas Mantissa claims to run x86 architecture on z hardware. Combine the two and you could run Windows on z/VM on a x86 emulating a z machine.Sounds snappy :-) BTW: MacOS did this quite successfully when they switched from 68000 to Power architecture. You could run your 68000 binaries and they would be dynamically translated into a cache Power instructions. They did it again when they moved to x86. See: http://knowledgerush.com/kr/encyclopedia/Mac_68K_emulator/ Of course, modern byte-code VMs use dynamic translation (JIT) to translate frequently used byte codes into native machine instructions. Yet, it is still amazing how many people think that Java is still only interpreted. Quoting from an Austin SHARE presentation on PHP: Unlike Java or other interpretive language, PHP’s focus is to use the script to invoke “native” C subroutines that do the actual work at full speed Kirk Wolf Dovetailed Technologies -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
I'm a bit curious. If I understand what is being done, it is similar to a Java JVM. The z/VOS reads and interprets the x86 arch instructions, with some JIT on the side. However, other companies, such as AMD, need cross licensing to create chips which run x86. That cross licensing is part of IBM's complaint against PSI. PSI was not licensed to create a zArchitecture emulator. Couldn't Intel shut down Mantissa with a similar suite? I don't remember reading that Mantissa has an agreement with Intel. I don't think the fact that z/VOS is a software emulator instead of a harware implementation is relevant. Of course, IANAL. -- John -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
IANALE, so wild speculation alert: Emulation *might* be different--PSI kept saying they didn't do emulation, although it quacked like a duck. The AMD licensing might be slightly different, since their architecture is slightly different. But it's a great question! I'd love to know the answer. P.S. z/Architecture. Slash for software (and the architecture, oddly enough, is considered software). On Tue, Mar 24, 2009 at 12:36 PM, John McKown joa...@swbell.net wrote: I'm a bit curious. If I understand what is being done, it is similar to a Java JVM. The z/VOS reads and interprets the x86 arch instructions, with some JIT on the side. However, other companies, such as AMD, need cross licensing to create chips which run x86. That cross licensing is part of IBM's complaint against PSI. PSI was not licensed to create a zArchitecture emulator. Couldn't Intel shut down Mantissa with a similar suite? I don't remember reading that Mantissa has an agreement with Intel. I don't think the fact that z/VOS is a software emulator instead of a harware implementation is relevant. Of course, IANAL. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
-Original Message- From: IBM Mainframe Discussion List [mailto:ibm-m...@bama.ua.edu] On Behalf Of Harry Wahl Sent: Tuesday, March 24, 2009 5:35 AM To: IBM-MAIN@bama.ua.edu Subject: Re: Old discussion about Windows running on a mainframe ( I brought up) A more sensible approach would be to look at creating a Windows HAL (hardware abstraction layer), or something conceptually similar, that runs on z/Series. Historically, this is how Windows has been made to run on different machine architectures. Of course, cooperative development between IBM and Microsoft would be necessary. Another possibility is to exploit the Infiniband feature of the z/10. This feature is profound in terms of 360-z/series evolution, but has been largely ignored, so far. Infiniband attached external hardware products that expose x86 architecture processors, from Intel or IBM itself (or maybe even Intel Larrabee), is an ideal way to run Windows on IBM mainframes. It seems to me that the obvious way to do this would be to build z86 on z in millicode. Then, the only real issue would be performance. An all millicode representation would be somewhat slower than the silicon. Dave Gibney Information Technology Services Washington State University -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
On Tue, 24 Mar 2009 11:36:42 -0500, John McKown wrote: I'm a bit curious. If I understand what is being done, it is similar to a Java JVM. The z/VOS reads and interprets the x86 arch instructions, with some JIT on the side. However, other companies, such as AMD, need cross licensing to create chips which run x86. Do they? That cross licensing is part of IBM's complaint against PSI. PSI was not licensed to create a zArchitecture emulator. I thought that the problem with PSI (and Hercules, for that matter) was not about the emulation of the hardware, but the licensing of z/OS. -- Tom Marchant -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
On Tue, 24 Mar 2009 12:49:26 -0500, Tom Marchant m42tom-ibmm...@yahoo.com wrote: On Tue, 24 Mar 2009 11:36:42 -0500, John McKown wrote: I'm a bit curious. If I understand what is being done, it is similar to a Java JVM. The z/VOS reads and interprets the x86 arch instructions, with some JIT on the side. However, other companies, such as AMD, need cross licensing to create chips which run x86. Do they? I am certain that AMD and Via have cross licensing agreements with Intel. I know because Intel has recently indicated that AMD is violating that agreement. http://www.amd.com/us-en/Weblets/0,,7832_12670_12686,00.html That cross licensing is part of IBM's complaint against PSI. PSI was not licensed to create a zArchitecture emulator. I thought that the problem with PSI (and Hercules, for that matter) was not about the emulation of the hardware, but the licensing of z/OS. Hum, parts of z/Architecture are patented. Not the hardware to perform the function, but the actual function itself - indepandant of the implementation. Neither Hercules/390 nor PSI had a license for those parts of the z/Architecture, even though they implemented them (not totally sure about Hercules/390 on this point). Also, as I remember (Danger, Will Robinson!), part of IBM's argument was that PSI's implementation of the z/Architecture was not verified by IBM. IBM said that in that case, the use of licensed IBM software (z/OS et al.) could not be guaranteed. However, much of IBM's premium for their software was due to IBM's guarantees on the reliability of the software. If the hardware was not up to snuff, then IBM might suffer damage to its reputation for reliability if the software (z/OS) were to fail. This was the logic as to why IBM refused to license z/OS et al. on the PSI solution. -- Tom Marchant -- John -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Old discussion about Windows running on a mainframe ( I brought up)
On Tue, 24 Mar 2009 13:55:38 -0500, John McKown wrote: On Tue, 24 Mar 2009 12:49:26 -0500, Tom Marchant wrote: On Tue, 24 Mar 2009 11:36:42 -0500, John McKown wrote: However, other companies, such as AMD, need cross licensing to create chips which run x86. Do they? I am certain that AMD and Via have cross licensing agreements with Intel. I know because Intel has recently indicated that AMD is violating that agreement. http://www.amd.com/us-en/Weblets/0,,7832_12670_12686,00.html The only thing I see in the page like that is this: In October 1991, Intel commenced a federal court action for copyright infringement. An arbitrator subsequently awarded AMD full rights to make and sell the Am386. The Supreme Court of California upheld this decision in 1994. Is that what you are talking about? That cross licensing is part of IBM's complaint against PSI. PSI was not licensed to create a zArchitecture emulator. I thought that the problem with PSI (and Hercules, for that matter) was not about the emulation of the hardware, but the licensing of z/OS. Also, as I remember (Danger, Will Robinson!), part of IBM's argument was that PSI's implementation of the z/Architecture was not verified by IBM. IBM said that in that case, the use of licensed IBM software (z/OS et al.) could not be guaranteed. However, much of IBM's premium for their software was due to IBM's guarantees on the reliability of the software. If the hardware was not up to snuff, then IBM might suffer damage to its reputation for reliability if the software (z/OS) were to fail. This was the logic as to why IBM refused to license z/OS et al. on the PSI solution. And one of the factors that makes the 360/370/z et. al. reliable is the machine check interruption. Does the Itanium processor have that capability? I'm pretty sure that none of the x86 processors do. -- Tom Marchant -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Old discussion about Windows running on a mainframe ( I brought up)
Will Big Blue mainframes run Windows? Track this topic Print story Post comment z/VMs get Microsoft rumor By Timothy Prickett Morgan • Get more from this author Posted in Servers, 23rd March 2009 22:34 GMT Whitepaper download - Eight CRM essentials An obscure mainframe software company called Mantissa Corporation bragged last summer on the IBM VM listserv - which is dedicated to virtual mainframe environments - that it was creating a product called z/VOS that would allow slices of a Windows operating system to run atop z/VM, the hypervisor-as-operating system for IBM mainframes. The product was due in the first quarter of this year, and the story of its impending release has been making the rounds. According to a report in NetworkWorld, Mantissa's z/VOS, presumably short for Virtual Operating System, is a layer of software for VM that allows desktop and server Windows operating systems to run in emulated mode atop z/VM. Mantissa - which is based in Birmingham, Alabama, and which is a supplier of report distribution and other tools for mainframes - talked about the z/VOS product at the SHARE mainframe user conference in early March in Austin, Texas. But that was not the same thing as a product launch. We've tried to reach the company for several days, but Mantissa has yet to respond. While IBM and the Linux community for mainframes centered around Marist College in New York have worked to get official mainframe ports done for Linux - Red Hat and Novell officially support mainframes, if you can write a big enough check to get support - there is no native Windows port to IBM mainframes as far as I know. So, the real curiosity is how Mantissa is supporting Windows XP or Vista atop z/VM partitions. According to the company's development blog, z/VOS includes a translation engine that converts native x86 code to its System z equivalent. See how easy that was? As it translates equivalent results - not creating equivalent machine code, mind you - the instruction that is created by z/VOS is stored in memory so it can be accessed the next time the operating system function inside Windows running on the mainframe is asked for again. Since Gary Dennis, Mantissa's chief executive officer and founder - and other we've called - are not answering their phones, it is a little hard to take the company seriously. But if it can indeed deliver a layer of abstraction software atop z/VM that lets Windows desktops and servers run on mainframe iron, the company should probably think about getting someone to answer the phones and maybe a salesperson or two to try to take some orders. If the x86 translation overhead is not too high, this could be a very interesting development - and one that Big Blue would seem pretty keen on supporting, not quashing. ® http://www.theregister.co.uk/2009/03/23/mantissa_windows_on_mainframes/ AT least these people aren't IBM types and IBM is not eager to dispel it (so I have heard) Ed -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html