With all the publicity that Windows and Linux get, you may be forgiven for not being aware of a number of other Operating Systems (OS). So in addition to my coverage OS/2 and BSD, there are some other choices that I find interesting and useful alternatives.
Since Windows, Linux and Mac are covered by other Byte writers, my emphasis will be on these alternative OSs.
Five Desktop Alternatives: * OS/2 Convenience Pack from IBM * eCS, OS/2 plus enhancements from Serenity Systems * Free BeOS, an open source version of the OS developed by Be * BeOS 5 personal * Amiga, the Phoenix of operating systems
Three Emulator/Virtual Machine Systems * VMWare * Virtual PC * Bochs
The last three OS in the list are qualitatively different from the rest - they are designed as emulators or virtual machines that can run more than one OS at a time, shared in one set of hardware. Like the story about the dancing bear, what is remarkable given the ancient x86 design is not how well they work, but that they work at all.
Be aware that this short list is by no means exhaustive. There are several more commercial or open OSs that are mature and in use, but even this list will be enough to digest at one time.
In the interest of full disclosure, I have used MS Windows from 1.01 in 1985 to 3.1, W95/98 and NT. In the 87-89 timeframe, I ran DesQview and Windows together. Since the 1991 beta of OS/2 2.0 through to today's eCS 1.0 and all versions between, my primary desktop has been on OS/2. In addition, currently I run NT, FreeBSD and OpenBSD as well as the occasional Linux.
People new to computers in the '90s probably don't know this background, and some may have forgotten. Microsoft was OS/2's chief competitor in the 90s, but it didn't start that way. Way back in the '86-87 time, OS/2 was originally a cooperative effort between Microsoft and IBM. Yep, the two goliaths were cooperating, in principle at least.
OS/2 became an IBM only project in 1989. Microsoft chose to develop Windows further in competition with OS/2. The cause of this battle is debatable, but it ultimately evolved into ownership of the desktop OS. The original Windows 4.0 was predicted for delivery in 1993, then 1994. It then became Windows 95 and was finally delivered in August '95 to the accompaniment of extraordinary hype and buildup.
OS/2 began as an enhanced OS to run protected mode programs, and multitask in the protected mode of the Intel 286 chip. The choice of the barely adequate 286 chip for this task caused significant development delays, so by the time it was available as OS/2 1.1, the 386 chip was already popular.
OS/2 2.0 became a 386 only version at IBM. It went beta in 1991, and shipped GA in 1992, just around the time of Windows 3.1 delivery. From early 1992 until August of 1995, Microsoft had no OS that could really compete in multitasking and reliability. Despite this 3 year lead in technology, IBM was unable to build a coordinated effort to sell OS/2, and one result is Microsoft has had an OS monopoly on the desktop for almost a decade.
The full story of this battle is much more complex and full of unusual events. For a variety of reasons, IBM had, in theory, the better team - more experience, better technical capabilities, a much larger sales force and entry into most business. So much for theory.
The situation has changed a lot since IBM's announcement in 2000 that OS/2 was end-of-life and would be supported with limited enhancements and drivers through 2006. Due to demand from a vocal business user base, IBM has increased its support to deliver upgraded Convenience Packs (CP) each year, reducing testing and upgrade support costs for the business community.
However, the strategy announcement for OS/2 for 2002 [http://www-3.ibm.com/software/os/warp/strategy/] contains some significant changes to the previous environment. Specifically:
"IBM does not intend to provide additional Convenience Packages in the future. For more information about Convenience Packages see announcement letter 200-082 at http://www.ibmlink.ibm.com/."
There's more bad news:
"# OS/2 Defect Support: Limited warranty defect support will expire for Warp Server for e-business on 31 May 2002 and for IBM WorkSpace On-Demand 2.0 on 31 December 2002. IBM plans to provide Program defect support for OS/2 Warp 4 Convenience Packages and for Warp Server for e-business Convenience Packages for customers with software subscriptions through 31 December 2004. "
It looks like end-of-life has been accelerated. The good news is next.
Equally vocal, but less financially convincing, were the individual users such as myself. Until 2001, it looked like CPs were my only option. Then another unusual event happened. IBM licensed OS/2 for resale to Serenity Systems, with the aim of supporting individuals and small businesses.
Serenity Systems enhanced their offering by improving the installation process, adding a large selection (35 items) of independently developed software and the Lotus Smart Suite from IBM as part of eCommStation (eCS), their version of OS/2. They also included SMP as an option for the workstation version.
eCS 1.0 shipped April 2001 and has continued development since then. Serenity Systems has indicated that eCS 1.1 is expected in fourth quarter of 2002 after some selected user testing. The list of enhancements planned is significant and includes a new installer, even better than the original eCS 1.0 installer, which was a big jump over IBM's version.
eCS site [http://www.ecomstation.com/] contains a lot of information - patches, new uploads, news, applications and links to other supporting sites. The support and FAQ [http://www.ecomstation.com/support-faq.phtml] section is particularly useful, containing items like this:
Several installable file systems are available for eComStation: * HPFS - the High Performance File System * CDFS - the CD-ROM File System * NETWKSTA - the LAN Server remote file system * HPOFS - the High Performance Optical File System * SRVIFS - The CID remote installation thin file system * JFS - Journal File System * FAT - File Allocation Table
It looks like Serenity Systems [Serenity-Systems.com] will be our future support for OS/2. Their performance in a tough business environment gives me confidence for the future of OS/2 in its eCS incarnation.
BeOS, is back live in two versions: The original BeOS 5 personal; And the new Open Source Free BeOS. Be was originally founded by a former Apple executive, Jean Luis Gastee. His vision was of an integrated hardware and OS that would be the best multimedia system ever. In a sense, it was like the birth of the Amiga except the industry landscape had changed.
Be began well. The early systems, built with dual processors as the standard, performed better at multimedia than any of the existing main players. However this was also a clear threat to both Apple and Microsoft, and both responded in typical ways. Apple improved their offering and Microsoft probably applied similar pressure to the recent demand that Dell ship an OS with every system.
Despite having a better multimedia mousetrap, somehow Be was never able to get a large manufacturer to step up to the plate to produce and sell globally or nationally. After an extended effort that probably would have succeeded in any other industry, Be was sold to Palm, Inc. Details of Be Inc.'s closing are here [http://www.beincorporated.com/]. I was among many who regretted that outcome.
However, all was not lost. While BeOS is still two or three resurrections short of Amiga's record, Be is here again in two forms - the original flavor BeOS 5 Personal, and the developing Free BeOS which is an Open Source project on Sourceforge.
BeOS 5 can be found here [http://www.BeBits.com]. BeBits is a layered web site for Be developers and users. It lists the latest developed software, lots of drivers, downloads and CDs of BeOS with many extras.
Free BeOS can be found here [http://open-beos.sourceforge.net/news.php]. [indent]
"OpenBeOS is a project dedicated to the re-creation, followed by the extension, of the BeOS.
Individual servers and APIs (known as kits) are being re-written from scratch by an enthusiastic team of volunteers who want to continue the revolution started by Be Inc. The kernel is being based on NewOS, a microkernel written by a former Be engineer, and adapted by a dedicated team of hard core programmers." [/indent]
The Phoenix of OSs, Commodore's Amiga, appeared in 1985 on custom hardware with a Motorola 68000 processor chip and some very advanced chips for sound and video. Amiga was first produced by Commodore Business Machines. Commodore restructured more than once, and was finally acquired by Gateway in 1997. It was acquired in December 1999 by Bill McEwen, who moved it to Snoqualmie Ridge in January 2000.
Currently, Bill McEwen is President and CEO of Amiga [http://www.amiga.com], now privately held. His products are called a "Digital Environment," which enables applications for devices from cellphones to large systems, running on multiple operating systems, or even as an operating system.
Despite multiple moves, the fans of Amiga, and its powerful video capabilities, kept bringing Amiga back. The Amiga's were video processors par excellence and could do image processing that was years in advance of other systems. One of the best known video applications of the Amiga was rendering the video images for Babylon 5, a science fiction TV show. A bank of 40 Amigas was used to develop beautiful and detailed imagery used in the SF series. The quality and smooth motion of the graphics is impressive.
The new Amiga is working on some very interesting products, which you can read about in an interview with Chief Technology Officer Fleecy Moss at TempleofTech [http://www.templeoftech.com/articles.cfm?ArticleID=58&PageID=1].
Amiga OS4 is still in development and OS5 is planned, but no schedule has been given. The official AmigaOS [http://os.amiga.com/os4/OS4Features.php] portal will give you up to date information on the status of this latest release.
Unlike many technology companies, they have held back on preannouncing specifics, but remain committed to delivering promised products when they are ready. It's a refreshingly novel approach.
Virtual or emulator machines form a small but important segment of the OS market. They work by either emulating the x86 instruction set (slow but general purpose) or by interpreting specific OS calls and connecting or translating them to the primary OS (faster, limited OS support).
VMWare [http://www.vmware.com/] can be hosted on Microsoft Windows NT 4.0, Windows2000, Windows XP, and Linux. The virtual machine supports the Windows systems from MS-DOS 6 and Windows 3.1 to Windows XP, plus the main Linux distributions and FreeBSD 3.x and 4.0 to 4.5.
VMWare identifies many usage scenarios, including: * Streamline development and testing (for multiple OS). * Support legacy applications migration. * Facilitate training and sales demos. * Improve help desk and technical support.
Of course the one that appeals to me the most is consolidation of six systems to one or two and still run software from any OS without having to get all my software for that one OS.
VMWare also has two server class virtual OS products. The GSX version is for standard servers and the ESX version is for servers with advanced hardware and workload management.
Virtual PC exists as two products - a well known Mac emulator for PC software and Virtual PC for Windows. Both are products of Connetrix [http://www.connectix.com/index_win.html]. In addition to the usual suspects, Windows and Linux, Virtual PC will host Netware and OS/2 on a Windows machine as well. They are the only VM to support a virtual OS/2 that I know currently.
At $229, Virtual PC is less expensive than VMWare Workstation at $299, but the choice should depend more on specific hardware and software support than price.
Bochs [http://bochs.sourceforge.net/] is an x86 emulator, and as such, will be slower than the other VMs listed here. It is being developed Open Source on Sourceforge and is available as a download for several machines. The source is available for people who are interested. While is is still in development, Bochs is functional for many uses.
For a review of these VMs, see this OSNews article [http://www.osnews.com/story.php?news_id=1054]. Also, check out an earlier report on VMWare [http://www.osnews.com/story.php?news_id=458].
Current virtual x86 machines are not efficient because the 386 architecture was not designed to virtualize ring zero protected mode instructions. What this means is operating systems, which use special instructions in ring zero, can be emulated only with great effort, and thus run slowly. This problem is caused by the 386 architecture and is not the result of inefficient programming.
This may now change. Based on my analysis of the AMD Hammer architecture and instructions, named x86-64, it appears possible for an x86-64 OS to virtualize the x86-32 protected ring zero instructions in an efficient manner.
If this analysis is accurate, then we may finally see a true Virtual Machine (VM) that can run current X86 OSs without a major performance hit. The implications of this are far reaching. If you currently use a multiboot system to run FreeBSD, Linux and Windows at different times, it would be possible to run them efficiently under VM-64.
If you run on separate machines as I do, you could consolidate into one larger machine and switch between virtual machines rather than physically turn switches to change keyboard, mouse and display connections.
Consolidation of my six systems into one large dual processor system would be a win for me in space, power, heat, performance and convenience. More than that, it would be a big win for any developer of OS or programs because they could be isolated in a separate copy of the OS. 32 bit OS crashes would not touch the VM, and even if the crash took the local OS down, the VM would still be able to explore memory to find the fault.
Restarts and full memory dumps would become a thing of the past. Multiple copies of Linux could run multiple Apache web servers to isolate production and test environments, and to separate company sites on a shared large machine. These things are done today on mainframes and large Unix systems from IBM, Sun and HP.
Extending this capability to microprocessor based computers could change the personal computer industry. Imagine being able to run your favorite programs, regardless of OS, on one system. Obviously the cost will include more memory and cycles, but the performance of the latest processors makes this easy even on a single processor systems. As a side benefit, even the memory manufacturers may finally be able to make consistent profits. [30]
All content on this site is Copyright 2001 and 2002 by Bill Nicholls