June 19, 2000
WarpTech2000 was put on by POSSI, the Phoenix OS/2 Society. This three-day technical conference was held May 26- 28, 2000 at The Wigwam Resort, a five-star resort about 30 miles from Phoenix, Arizona. The conference was memorable for people, atmosphere, and content. A list of the sessions at www.warptech.org/WTSessions.html gives a picture of the range of topics this ambitious conference covered. From Remote Booting Alien Operating Systems under OS/2 to Connecting Databases to Java Applications, the sessions covered a broad range of technology and how to use it with OS/2.
Leading off the second day of the conference was Steven King from IBM. In an extended presentation, he explained IBM's plan for future OS/2 support. In broad terms, IBM plans e-business enhancements to OS/2 through 2002, device driver and defect support through 2004, and fee-based enhancements and support through 2006. In my interview with Steven, I asked if this had been approved by John Soyring, vice president e-business operating systems solutions (e-boss).
His reply was that it had been reviewed and approved all the way up to Lou Gerstner, CEO, and was now an IBM commitment. King also commented that some customers had been requesting commitments through 2008, though those were not yet approved.
King talked about the financial performance of OS/2. Both 1999 and 1Q2000 were well above the advanced plan. King builds a base plan for revenue and an advanced plan for a higher revenue stream. He stated that the company had achieved 'millions of dollars' over the advanced plan. He also said that starting this year, he can apply the higher revenue to new developments in the following quarter, rather than having to wait until the next year. All of this will accelerate work on OS/2 support.
Planned enhancements for Java include skipping Java 1.2, jumping to Java 1.3 targeted for July 2000. Browser enhancements, scheduled for 1Q2001, will be based on Mozilla, the open source version of Netscape and will include:
Another focus is enabling e-business. IBM encourages users to implement solutions based on standards like TCP/IP, Java, and XML, with a browser for user interface, and Domino and WebSphere application servers for application hosting.
Other steps planned for 2000 include enhanced device-driver support:
Technical assistance to manufacturers, OEMs, IHVs and ISVs LAN Adapters ~ 15 per quarter USB support via device-drive package IDE and RAID for IBM hardware SCSI by OEMs with technical assistance from IBM SciTech GRADD video device drivers New printer device drivers ~ 20 per quarter DVD and UDF file support and bootable CD-ROMs for Warp 4
Convenience packages are a single CD with Warp 4 or Warp Server for e-business (WSeB) that include new features like the previously listed enhancements, software fixes, device drivers, and other possible additions. When I asked King who originated this idea, he said "The customers." While it isn't obvious to most of us, maintaining OS/2 installations has become complicated. Any new machine requires an original install, the latest fixpak installed, device drivers updated, and possible software patches applied. Then you can start testing applications.
It's complicated even with just the four OS/2 systems I have. Now, consider a large commercial installation with 1,000 OS/2 desktops. Every time a fixpak comes out, that company has to spend resources to test the fixpak against its applications in its environment before upgrading any user systems. By the time one fixpak has been tested and installed, another fixpak is out. Thus, these companies have an ongoing requirement for testing and support that can be quite expensive. King quoted typical testing budgets of $250,000 per year.
IBM's convenience package addresses this problem. Once a year, IBM will release a fully tested OS/2 package with a common code base and all current enhancements and patches applied. Testing will be a full-court press from IBM, which tests the package as though it were a new release. Once this is done, customers will not need to assemble their own set of patches and upgrades, and the extensive IBM testing should reduce the customer testing to a minimum.
Customers win from only one major change per year, with IBM paying for testing, so customers need to spend less. Stability will improve and bug fixes will cost less for IBM since there will be a standard platform. Buying the convenience package also extends the OS/2 support 12 months from the date of delivery.
Selecting how you get the convenience package is a little complex. The first requirement is a Warp 4 license. If you are pre Warp 4, upgrading to that is the first step. With a Warp 4 license, you may choose to buy the convenience package as a unit, subscribe to a software choice that will include the package, or earn the convenience package with Passport points as an IBM customer.
Once you have decided how to get the package, you can select how many years you want to subscribe. King has pointed out that it is more expensive to drop out of the subscription and have to rejoin than if you stay in continuously. These various options provide IBM with a stream of revenue to support OS/2. One of the key points to come from this presentation is that ongoing support for OS/2 will not be free. Once the Warp 4 product reaches its end of service life, any further support will require one of the support options above, or a custom support contract with IBM for specific requirements.
Given that IBM is a for-profit company, none of this is surprising. King made it clear that IBM customers wanted solutions to business needs, not raw technology. From a corporate viewpoint, technology is supposed to help solve business problems faster and easier. This is sometimes hard to grasp as a SOHO user with just one or two systems, using standard shrink-wrap software.
Remember 1992, when OS/2 2.0 arrived? A hot choice that year was a 486/33 with <gasp> 16 Mbytes of RAM and a huge 120 Mbyte-hard disk; 14.4 kbps modems were new, bulletin boards were growing and almost nobody had heard of something called the Internet. Just a few physicists who wanted to share data and programs between the major research sites built the basic Internet infrastructure. Nobody then had a clue as to where it would go.
The dramatic changes in the past eight years have not changed the essential nature of an OS. Without it, a computer is an expensive, electrically powered doorstop. What it has changed is the role the OS plays in applications. No longer does the OS completely define what applications can run. Once standards are supported, any application that talks to a browser is available on any OS platform. That is the reason Microsoft struck out at Netscape and Java. If those two standards catch on, it won't matter what OS you run, and Microsoft's monopoly goes poof.
In the opinion of many users, IBM has given us the best desktop OS in the past decade, yet it is necessarily responsive to the large customers who generate revenues measured in six figures and more. For eight years, OS/2 SOHO users have had a free ride on the tailgate of these large customers, while the whole environment changed. In this new environment, communications and standards are king, the browser is the queen, and the OS has taken a supporting role.
To many people in the corporate environment, the desktop OS is no longer an issue except for cost. The last part of King's presentation showed the future path for OS/2: Prepare for the application framework for e-business. Deploy the application framework for e-business. Make the transition of the client operating system. Where OS/2 fits in this future is as an enabler for the transition to treating thin clients with browsers differently from fat clients with big OS and applications.
More than half of the future desktops will be thin clients, where all access to applications will be via browser. Even fat clients will be remotely managed for cost-control reasons. IBM proposes Workspace On Demand (WSOD), OS/2's fraternal twin, for these fat clients. An embedded OS/2 with a browser might even be possible for thin clients. In any case, the computing world has changed, and operating systems will change as a result.
While some people know Odin is the Win 32-OS2 project extended, many are not familiar with its objective. Odin is a set of program elements that provide execution compatibility for Windows software, Windows 95, 98, and NT, executing as native programs under Warp. This is a huge technical challenge involving everything from an incompatible loader format to thousands of API calls, many undocumented by Microsoft.
According to the speakers, IE5 alone has over 2,000 API calls, of which only 500 are documented. The presentation, given by Patrick Haller and Achim Hasenmller was casual and low key, the reaction anything but. It had the complete attention of an SRO crowd, intense silence punctuated by bursts of applause and laughter. One of the biggest laughs came when someone asked if Outlook would run under Odin. The response was, "You mean the love bug accelerator?"
The slide presentation gave a detailed history of the project, delved into Windows and OS/2 architectures and dealt with some of the challenges they faced and overcame. One interesting point was the Odin project was designed to run any Windows program from the Win 95, Win 98, or NT 4 group, including those that Windows restricted to Win 9x or NT only.
The Odin project is roughly equivalent to the challenge of climbing Mt. Everest without bottled oxygen and with one eye closed. Odin is at camp six, getting ready for the final assault. There are many Win 32 programs already running, not just games as was my earlier impression.
Being a programmer with OS experience, I was stunned by what this group of eight(!) programmers had accomplished in their spare time. Odin's history started as the Win 32-OS2 project in May 1997 with Sander van Leeuwin experiments running W95 games under OS/2. In August, he was joined by Peter Fitzsimmons, Vince Vielhaber, Joel Troster, and Felix Mascheck. The vision moved from games to general Win 95 support. In January 1998, Patrick Haller and Knut Osmundsum joined the team.
That year, they released four alpha versions that were capable of running Quake 2, a very popular game. In 1999, the project was renamed Odin and moved to open source supported by Netlabs. Jumping to the current time, Odin now supports most Win 32 APIs with new windowing management and a ring 3 loader. There are no fundamental limitations remaining, including programs that load beyond 512-Mbyte virtual space.
This limit was removed with the release of Aurora and Warp 4 with FixPak 13, allowing access to a full 3 Gbytes of virtual space. The current release of Odin runs the following Win 32 applications well: Realplayer2 Adobe Distiller Notes 5 Framemaker 5.5 Quake MS Word97 running on a Warp 4 system was unveiled at WT2K. Note that some programs may yet have glitches as Odin remains a work in progress.
To see what runs and how well, check out NetLabs on the Web, click on the applications under Misc., and use the search capability to see which applications are rated operational.
I've seen a lot of presentations. They range from the absurd to the valuable, more lightweight than content heavy. Most of the time, the audience leaves with some relief as they slowly get moving again. The Odin presentation was unique in my experience. I didn't notice the time pass, never had my attention wander, and was as disappointed as others were when it was over. Patrick and Achim could have read from the telephone directory without losing anyone. At the end, they got a standing ovation that lasted for minutes. It was richly deserved.
The most interesting speaker I met at WT2K was Kimwai (Kim) Cheung. Kim is a retired transmission engineer, electrical transmission, as in the 500 and 750 KV main transmission lines that distribute power through the United States. At dinner Friday night, he talked about his former job and the changes in the U.S. power system since he retired. During that time, spinning reserve dropped from 12 percent to 5 percent of power demand, and Kim wonders why we don't have more power failures. From what he talked about, it sounded like investing in a company that makes auxiliary power generators might be a good move.
This is not exactly the picture of your typical marketing director. Kim has originated and brought three products to market through Serenity Systems. They are Wise Talker, Wise Attendant, and Wise Manager. The first two are telephone-answering software tools designed to make building custom telephone systems a simple process. Wise Talker is for voice trees and response, while the Attendant is more powerful and can route calls and build custom-calling systems. His software supports the four major suppliers of professional telephone cards.
Wise Manager is a program of another class entirely. Wise Manager runs on WSeB and enables remote loading of any OS from the RIPL server to any system, including thin clients without boot proms or any disk. His demonstration was more impressive than most, as an unplanned power drop hit the entire string of systems he was using as a demo, including the server, while he was loading a Win 95 system to a thin client. With power restored, the server recovered, other systems came up, and the thin client reloaded the Win 95 system without special recovery.
Having demonstrated a robust product, he went on to the icing on the cake. The real benefit of the Wise Manager is in applications management. Wise Manager can extend to multiple zones on a LAN, with each one being independently defined and managed. Adding an application to a single system, group of systems, or an entire zone was almost too simple to believe. He dragged the application icon to the systems or zone that he wanted to install, and it installed. Demonstrating this on the thin client showed how quick and easy it was.
Compared to WSOD it seems too simple to work, but it did. Want several programs installed? Drag the whole bunch. Updates? Ditto. Even setting up the parameters for the install was automated. On a test system, install the new program manually. Wise Manager captures a before and after image of the files, dlls, registry entries and config.sys changes, and stores the changes for automatic installs -- or for that matter, uninstalls.
While this is not an inexpensive program, starting at $795 for 5 clients and one server, the payoff in terms of reduced support and deployment costs makes this a less than 12-month payback, quite possibly less than 6 months. For larger environments, once installed, payback could be even faster. This is not including the side benefits of having consistent software on all stations and easy backsteps if an upgraded program has serious problems.
I'm looking forward to giving this a hands-on test on my own network, and will report my own experiences. For any of the Wise products, check with Serenity Systems in the link above.
Along with logical names like Wise Manager, how about Simplicity for Java? OK, I can hear you complain, it can't be that simple. That's what I thought. Demonstrating Simplicity Pro from Data Representations was the chief programmer, so I wasn't surprised at his ease of use. Obviously he knew the whole system cold. I was prepared for the usual demo of instant coding tricks, prewritten code for the demo, all the usual shortcuts.
He didn't use any of them. Not only that, he didn't write one line of code. None. For each step in the process, he pulled down a menu, selected a choice and was led through a series of English-language choices as to what that piece should perform. Upon finishing each menu item, the generated code was instantly available and instantly able to run in an application window. In fact, that window showed the current state of the application as each piece was dropped or selected.
None of the connections required by Visual Age for Java, or anything like that, was ever needed. As I watched the process, I became aware that each piece of the code knew how it related and where it fit in the system. In demo mode, it took about 10 minutes to show how to build a simple data-display application, including explaining what each step did. That was impressive, but it was only the warm-up.
Next, he said "Let's attach this to a database." I figured this is where the shortcuts would come in. Wrong again. Using the same technique, he selected a JDBC interface component, added the DB name and selected a table from the DB. Setting up a four-field display, he proceeded to add DB inquiry and display capability in the next five minutes or so. Which DB? Any DB with a JDBC or ODBC interface, or even a flat file.
He said for a few thousand items, a flat file was fine. An Excel spreadsheet would work too. Along with hooking up the database, Simplicity Pro wrote the SQL code for the data retrieval, and could modify it for the custom inquiry he added on after the first DB step.
My "seeing" was keeping up with the demo, but my "believing" was back somewhere over Mississippi. I was having intellectual jet lag trying to absorb all the implications of what the tool could do. I've been a programmer a long time, and complained in a recent column that programmer's tools hadn't advanced very far. This one looks like a real quantum leap. Talking about the product's productivity, he compared a three-week job in C++ to being done in a few hours in Simplicity Pro.
Plus, it runs on any Java system, with any database. Do you want fries with that? I was going into overload. He wasn't finished. Simplicity is written in 100 percent pure Java and runs on any compliant Java system and generates 100 percent pure Java code. The demonstration had been run on an old laptop with a Pentium 150 and 48 Mbytes of RAM. It was snappier than Visual Age for Java in 64 Mbytes on a Cyrix 300, which I have.
So, what's the downside? I can't find one. Despite a $795 price tag for the Pro version, which supports database access, this isn't hard to justify with the power of the tool. There is also a $195 version named Simplicity for Java that has all of the features except the DB support. And there is a Simplicity Pro demo for download on its website. So, what are you sitting there for? Click on the link and check it out. I'm awaiting a copy to test for myself and will report on my experiences later.
Servers are ever more important to the operation of a company today, and this will only increase. Downtime at $100 to $10,000 per minute is simply not acceptable. Loss of data or customer records causes excruciating pain for the IT manager who lets that happen. Thumbscrews may come back in fashion if IT can't get things running reliably.
Fortunately for Warp Server, there's Co-Standby. Formerly Vinca, it has been rewritten and renamed under a new owner. Installed in a pair of servers, Co-Standby lets the manager set up different processes on each server, which run with data mirrored to the other server via a dedicated Ethernet connection, even across town or across the Internet. These processes may be for different services or the same, but the same services must split the supported LAN address space.
Both servers can be actively in use with that one limitation. Upon detection of a failure or manual override, prewritten scripts shut down a process and start it up on the other server. This is particularly useful for upgrades of software or hardware. During a low-utilization time, move all the processes to one server, shut the other down, and upgrade it. When done, power up the server and redistribute the processes between the two systems.
The users won't notice anything except possibly a temporary performance decrease. Obviously, that's a simplified description of the capabilities. What it means is that, except for unusual power failures, 100 percent uptime is achievable. As a bonus, most of the upgrade constraints are removed with the ability to switch the load to either machine. Even a forklift upgrade is possible without having a full system-down condition. Since the system will rebuild mirrors across to the new hardware, all you need is Warp Server and Co-Standby installed before switching hardware.
Currently, Warp Server for e-business support is under development and I don't have a scheduled availability date. But it is a planned product. For the latest info, check out Legato.
WarpTech2000 was an outstanding conference. My only regret, besides the Arizona temperatures, is that not being a twin, I could only get to some of the sessions and had to miss others of equal interest. This one is a keeper; don't miss it. To check out the full list of sessions, click on POSSI or the WarpTech link.
All content on this site is Copyright 2001 by Bill Nicholls