miércoles, julio 07, 2010

El contínuo ser o no ser de IBM con el AS400

Bob Cozzi, veterano experto en el AS400/iSeries/System i, comenta la errática historia de soporte del AS400 por parte de IBM, concentrándose en el punto que constituyera su peor decisión estratégica sobre el impecable AS400: no desarrollar una presentación gráfica, ni salir jamás al cruce de quienes pusieran el acento sobre este aspecto para discutir de "modernidad":

Bill Gates & Co. were right: graphical user interfaces (GUIs) are "better." Today we know that in the mid-1990s IBM Rochester had the opportunity to move the operating system to a graphical-based user interface. This was happening around the same time as the move of OS/400 to the PowerPC chipset. This move gave OS/400 the platform the performance boost it so badly needed. But for some reason, IBM decided against a GUI for the i platform. This wasn't a simple "thumbs up/down" vote, however. I have heard it was a heated debate that came down to this: "our customers have so much invested in green screens it will be difficult to get them to change."

Of course this was nothing more than the classic: "we've always done it this way so why change now?" I have also heard there was concern about performance. It would have taken a lot to put a native GUI on the platform, but would it perform? If they created dumb-head terminals, (which they did) and allowed a GUI to run on those (which they did) and then allow them to be attached to the i platform (which they did), then the GUI engine could be outboard (off the main system) and therefore not impact performance (which they didn't do). Two retired IBMer's (the same ones who built the original 5250 data stream in the 1970s) independently leveraged all that technology IBM created and didn't use and enable support for a native GUI on the i platform. This happened at least 12 years ago—perhaps more—and it worked great. But on the i platform, unless IBM gets behind a technology (they did not) rarely does it have long-term success. So the two former IBMers stopped selling their GUI just a few years ago.

If IBM had created a GUI and advocated the platform as an end-user product with a server option (like everyone else seems to be doing), then it could have been i from the end-user desktop to the server farm and everything in between. The IBM POWER-based Cell chips used in Nintendo and other video gaming consoles make graphics scream—more than enough power for OS i GUI support. Imagine the graphical performance boost we would have had when Cell came out and was added to the i platform.

It was Lou Gerstner, one of the most respected IBM CEO's of all time, who, in my opinion, made one of the worst decisions any IT industry CEO of all time. What is it? Gerstner said "When something becomes a commodity we won't be in that business." Well, Lou, today all IT hardware is "a commodity," so now what?

Pero Cozzi no habla de esta ya demasiado discutida cuestión porque sí. Lo hace porque sin duda este paradigma a devenido igualmente "obsoleto", y la arquitectura que ahora generalizadamente es discutida como más idonea está basada en la web. Su pregunta es si ahora IBM está dispuesta a explotar este cambio, que por otra parte está disponible en el AS400 desde hace mucho tiempo (la primera vez que abandoné un Internet Information Server en favor del básico HTTP server del AS400 sucedió hace alrededor de diez años).
Sin embargo, del tono de su argumentación se desprende algo adicional: el problema con el cambio no sólo proviene de IBM, sino de su base de usuarios, y de su nivel gerencial de usuarios. Sólo un cambio político de IBM podría generar otra persepción del equipo y sus posibilidades:

Today, GUI means browser-based user interfaces. You can build GUI applications for i (iGUI as I call them) using RPG, HTML, JavaScript, and PHP, so there is no excuse to allow those Microsoft weenies into your shop to take away your job.

The i is so reliable and trusted over other platforms that it's now being installed all over the world in countries such as Russia, where industries like banking or gas and oil are fast-growing industries. Those systems and applications need to be not only reliable, but secure—secure not just from outside attacks/hacks but from untrustworthy in-house employees. Recently I visited several banks in Russia and I saw i running in these banks. When this kind of security is required, nobody screws around with Windows, and Linux is just too labor intensive to be used efficiently.

Remember, when you have tens of millions, or hundreds of millions, or in some cases, billions of records in the database, those other platforms that a Microsoft-weenie boss might be using at home on his 12-record checkbook application just don't scale up for use in a real-world environment. If he or she is impressed with a sexy WYSIWYG GUI application they use at home, it is your job to let them know the i platform and RPG IV also support GUI applications.

When your company wants a new application, strongly suggest that it be written with a browser-based user interface.

When your company refuses based on "all the other applications are green screen", you must refuse their "we've always done it this way" argument and push for an iGUI solution.

If your company still refuses, suggest that they allow you to write the new application in both green screen and iGUI versions—this will let them see what the i is capable of doing and give you the much needed GUI experience. Remember, you're first five to 10 web-based iGUI applications may not be very user-friendly; you are a programmer, after all.

When your company still refuses, then write the green-screen application and, covertly (perhaps on your own time) write the same application using an iGUI interface. Then when you roll out the app, suggest that they take a look at the browser-based version as well.

If they still refuse, then accept it and continue this practice for the next four or five applications. Then when the new VP of IT is hired and says "green screens are old, we're going to Microsoft" you can say "we've got GUI versions of these applications already". So the "i" suddenly doesn't look so old.

If they still refuse, at least you'll have some web browser-based, iGUI application experience to add to your resume.

Esta larga historia de olvido del desarrollo de una interfase de presentación gráfica ha sido siempre reemplazada por terceros: en primer lugar, al margen de IBM, que probablemente haya perdido un volúmen de ventas incalculable. Y, en el marco del mercado del AS400, por el desarrollo de herramientas de terceros, en general de generación automática de código, o de desarrollo basado en modelos. Dentro de IBM, por su compra de Rational; fuera, por productos como Plex, que naciera con dos propósitos: generar una interfase gráfica para su esquema ya existente de generación de código para el AS400 (2E), y para trascender este mercado, y generalizar el desarrollo basado en modelos a arquitecturas abiertas.

No hay comentarios.: