Saving the System i: Fight Pervasive with Pervasive
December 11, 2006 Brian Kelly
In a few years, there will be nobody sitting at a desktop or laptop PC anywhere in the world whose machine bears the letters “IBM.” Can this really be good for IBM? Does the principle of “out of sight, out of mind” jump out at anyone? The information technology adepts (fondly referred to by peers as “geeks” and “nerds”) who run Windows, Unix, and Linux shops on non-Blue systems will see no little letters “IBM” in the whole of their organizations.
This would not simply be because there are no IBM servers, X64 or otherwise, but because there are no IBM desktop or laptop PCs. IBM seems to be more than content to resign itself to this impending twilight reality.
The client PC was obviously an albatross for the IBM, which Lenovo removed from around its neck. The problem was that it was the same type of unit that a regular non-IT consumer could buy and IBM was just not in the consumer business market. When has popular consciousness ever mattered? Always. But frankly, with IBM’s propensity to spend little to nothing on consumer advertising, Big Blue had no real way of talking with John Q Public. With JQP being the biggest buyer of PCs, IBM had little influence on PC decisions. And so, when its client PC market share began to approach 5 percent, IBM sold to Lenovo and, according to senior management, IBM is not about to revisit that decision anytime soon.
In businesses and in homes, therefore, it will be a long time before any of these folks, buried in their non-IBM platforms or their game machines, will see or hear anything IBM. Even the loyal soldier in the AS/400, iSeries, and System i5 shops in the world will have no IBM on the desktop or the laptop. And when these fine IBM people go home, the same will be true. No IBM.
With or without IBM, the tens and even hundreds of millions of people who run Windows at home and also in the office every day will continue to think that Windows is an acceptable client computing platform. With the introduction of Windows 2000 and then with Windows XP, the crashes are substantially fewer and the irritation level is not as high–even for professionals whose IBM systems seem to never fail. So time has been good to Microsoft and Intel–and now, even Advanced Micro Devices–since today there is no question that well behaved software does not crash the operating system as often. Thus, the good ole blue screen of death is no longer a constant reminder of how bad it is. Though it may not yet even approach the robustness and reliability of the mighty IBM systems, the Windows client has become very amenable to desktop computing–except for those of us that long for something else.
Ah. Machs Nichs! IBM doesn’t sell PCs anymore anyway. Don’t worry. Nobody is suggesting that IBM re-enter the PC business as we know it. However, there is some room for something along the lines of a top quality machine for developers that could be a means of introducing the marvels of IBM systems to a developer who simply cannot afford his or her own System i5.
If the idea that I am about to explain is adopted by IBM, the slippery slope theory does suggest that the System i Developer Workstation could theoretically become something much more. It might start off as a System i with a Power Linux desktop, but it may evolve to become a mom and pop server, taking care of five users, to a powerful Xbox simulation machine. It may even become a PC that behaves almost like a Wintel unit. But, whatever it becomes it will get there because it can provide the most modern intelligent development environment that IBM could imagine. And we all know how much IBM wants its System i developers to use WebSphere Developer Studio. There’s no better way than to have it run on Power on a partitioned box that also provides System i server capabilities.
In the spirit of reusability, the part of the machine that would be used for the i5/OS partition could also be the AIX partition for a Unix developer. Likewise, it could be a Linux partition for the Linux developer or a z/OS partition for a mainframe developer (when and if the Power chip begins to support the mainframe). If or when Windows is ever brought over to run on the Power platform, this little machine could also have a Windows partition to support the Windows developer. It’s pretty obvious what the next step is: The All-Everything Desktop Development Client that comes with a multicore chip and supports development for all operating systems on the one developer station concurrently. It can go even further. There’s room for Solaris and Mac OS and any other OS du jour that needs residency on the most capable development box ever conceived. And, if Big Blue chooses to do it, this box will have an IBM logo.
We’re getting ahead of ourselves here, though. Let me first tell you what this machine is all about, its primary role as the System i developer machine. If it can be built to just do that, it will be impressive indeed and it would stand ready to help IBM and System i shops in many specific ways.
I must reiterate that I am not suggesting the System i division create a PC or a five-user mom-and-pop server, or an all-everything developer machine or a “WOW!” game box. However, I am suggesting that all of these things are possible if the System i Developer Workstation is as successful as it ought to be.
As I sit at my Windows XP “word processing” machine right now, I can access most of the System i units that reside in my clients’ shops as needed. I can also access the academic applications running on the System i at Marywood University, where I work, over the Internet via iSeries Access or Navigator or other means. I can also invoke the big client version of the WDSc software. I can emulate some WebSphere stuff or edit a source file or WebFace an application right on my own PC. Of course, I can also use the browser to test some Web code in PHP or WebSphere. I can get emails from Domino on the System i telling me what a great job I am doing. I’ve got a lot sitting on my desktop for sure and I really should be happy because there are some days that this Windows XP white box, no matter how hard I tax it, does not go lock or go down. Then of course, there are other days. . . .
However, with all my desktop and wide-as-the-Internet-is-deep power, I have no System i of my own. Just like you, or your home office, or each developer in a System i shop, I cannot afford today’s $20,000-plus entry System I machine. It’s just too expensive. I cannot afford one nor could any shop afford any model System i today for each developer in the shop. That does not stop me from dreaming. If there were such a client machine available to me as a developer, researcher, and solver of problems, I would be willing to pay more than twice the cost of a PC for such a machine–in the neighborhood of $2,000, and maybe even more.
Though it is not as bad as it once was, my Windows XP machine still locks occasionally, forces me to lose my desktop data, and causes me to endure its less-than-delightful reboot process. As a developer and consultant myself, I would love to have a desktop System i with the client side and the server side on one Power-based machine. Wouldn’t you?
The machine as I envision it would already come packaged with all I would need. It would have an invisible (in terms of work) desktop Linux partition and an i5/OS partition. When you boot the machine, the Linux side would come up with a handsome Gnome or KDE desktop that has a look and feel similar to Windows, which stole the idea from X Windows anyway. It would have icons for all of the client side System i software that I would need to use the box as a bona fide development machine. None of this has to be invented. This stuff all runs today on any number of desktop Linux distributions. This user-friendliness caters to the wide mass consumer base.
Then, all of the utilities that a modern developer would need for a System i, like iSeries Access, and WDSc, would be on the Linux side. The system side in the second partition would run i5/OS and it would have all of the facilities and all of the work management available on any production system. This would include the base i5/OS functionality as well as the Apache Web server, the new internal Web application server, and WebSphere Express in both its hidden and materialized forms. Don’t you just like it already? How convenient that it would also sit snugly on a desktop.
I would think most System i developers and consultants are not like me, writing magazine articles and books in their spare time, so they would not necessarily need all of the additional client tools that I need. But in my idea, nobody is left out. Given IBM’s estrangement from the public eye, over-inclusion is a self-evidently better theme here than exclusion. Who would want a Windows machine accompanying their developer’s machine so they could do the typical PC chores like word processing and spreadsheet crunching? I would need the tools to enable me to write articles and books. So to keep me from needing a Windows client machine to do my client stuff; and basically to keep the desktop footprint to one unit, Linux client applications need to be added to the total developer’s package. And the desktop version of Linux today is mature enough to handle even this task for most, if not all developers.
For example, everybody needs email to exist, so a Linux e-mail client would have to be on the machine in addition to the development tools. Most people also need spreadsheet, word processing, and presentation (Powerpoint) capabilities. All of these are provided by the OpenOffice suite, and they are downloadable for free. For instance, anyone can venture out to OpenOffice now and download the software for Windows PCs. It works in both environments, and on Solaris, Mac OS, and FreeBSD, too.
Just to see how robust these “PC-type” Linux applications really are, I conducted my own test on my Windows XP. I used the OpenOffice word processor (called Writer) to read the .doc file and then I proceeded to edit two of my books, The All-Everything Machine and Chip Wars just to see if it could really cut the mustard. It actually does perform up to expectations and perhaps can even go further. In all candor, I can hardly tell the difference. I freely admit that I continue to use Word in book development because I am not yet fully comfortable with trusting of the OpenOffice ware, and there is the potential of a lot of redundant work in assembling a book. But I am hoping to switch at some point since all of my testing experiences have left me quite pleased. My perspective is that this is already good enough for the typical developer who is an occasional PC user.
I have another experience to relate. My one son who is in law school panicked one day when his PC crashed. He bought another but it did not have Word on it. He did not want to pay for it, yet he could not retrieve his own Word to make his new PC complete. Over the phone, I had him download OpenOffice and to this day (one year later) he continues to use the Writer from OpenOffice and he submits his briefs in .doc Word format and they are accepted.
To have any credibility, I knew I had to do more for this article than tell you that my son uses it and that I was able to do some innocuous things with it to several precious and large files (my books). To put my money where my mouth is, however I wrote this article with Writer and I bet you couldn’t tell. So far, at least, I can’t either.
For the database function, the Openoffice Base program provides similar function to MS Access, but it is an ongoing in-process project and is not as mature as the other free Office-alike applications in the package. The open source movement is very strong, however and apparently intent on denying Microsoft future revenues on the Windows and Linux platforms. So, I would expect that it will come of age and fairly quickly. For those who want more than a desktop database however, there is the reasonably powerful MySQL database which almost all real geeks like much better than MS Access anyway. And, of course, there is the opportunity for IBM to load its client DB2 package, or even the freebie DB2 Express-C, on the Linux desktop to give System i an IBM client level database package.
IBM can do even better than this to create an impressive development machine for System i. For example, the IBM Software Group could preload all of its products built for System i and provide them for free on the i5/OS side of the development station. This free software would not be trial ware but could actually be used for development purposes free of charge without the need for disrupting the major production System i unit. I’ll get back to this.
There are a lot of other good reasons to have a System i preferred development workstation, regardless of how it gets nicknamed. Perhaps Little i and the production model gets nicknamed Big i–or perhaps not. Quite frankly, I haven’t given much thought to the name but the simple ring of the System i Development Workstation sounds pretty good right now. It is what it is. This new workstation box is a good idea for IBM for many reasons, not the least of which would be to have an affordable box that can be used anywhere by anyone, not just the privileged, for all types of productive purposes such as development, learning, testing, and single station productive work.
How many of these Little i boxes do you think could be sold? With hundreds of thousands of System i developers across the world, the word of the success that one developer would be enjoying on just one desktop in one System i shop would spread quickly. In the same shop every developer would then want one. Other shops would hear about it. We’d be writing about it. COMMON would be offering them as prizes. Even the accounts payable clerk would discover it as would the executives in System i shops. Perhaps as many as a million of these Little i machines could be sold in a short period and everybody is a winner. Then who knows how far this idea could go. Maybe IBM would re-announce the PC again under Power and the “P” would stand for professional or maybe even “Power.”
Then, for you GNU licensing buffs, Power as in the Power PC would be recursive. GNU, as many of you know, is a project to build a Unix-like operating system, the kernel of which today is Linux. GNU is a recursive acronym for “GNU’s Not Unix.” It is pronounced “guh-noo,” like canoe. To make the Power PC (not PowerPC as in PowerPC chip) recursive, however, a few English gurus would have to work on it, but it’s a good idea and it might help us all avoid a world in which only Windows and Intel have any say. Windows and Intel are recursive enough for my English.
Anyway, dream about a developer running a partitioned Linux /i5/OS desktop with a built-in hypervisor and the virtualization engine to match the latest OS release. Should that not automatically translate into a removal of the fear of putting on a new production release? Should that not relax the fear of working with partitioned systems? Actually, with so many IBM would definitely have enough cases to get it right. So, helping System i shops stay current with the OS is yet another reason for the workstation. There are way too many System i shops that are back-level on their releases. Many of these actually are by design. It’s the “no-hassle” approach to systems management: Never upgrade by design. Many are nowhere close to V5R3. I know that System i folks or AS/400 folks would argue that it’s because they are busy, but sorry, that’s not really a good excuse. It may be true but that doesn’t make it good.
So, this Little i becomes a staging platform for new releases plus test platform for new software that just might fit. Doesn’t this sound like the perfect pair of other uses for the System i Developer Workstation? If all of the whiz software division products that now sell very slowly on System i are installed on the i server side, a “try it you’ll like it” offering can become such a potential money maker for IBM that the Software Division may even be prompted to give these little guys away just to sell more software to System i shops. Today, mostly because they are overpriced for the System i market (and also because nobody really understands what they do), the IBM software division products rarely find their way to the loading dock, let alone the production System i. Additionally, nobody would argue that there is a lot of red tape involved in testing a product on a machine being used flawlessly by many potentially angry users.
There is really a lot of work in putting an untried package on a production box but none of that work is equal to the amount of work it would take to order even a test version of a product. So, it would be nice and easy to have System i IBM software products of all kinds on every one of the developers boxes in every shop. Chances are that someday, sometime, some developer would give a given product a try, find out it has value, and suggest that it be purchased and be moved to the big box. The Little i can sell IBM software for sure.
This super development machine, as intended especially for System i professionals, can be used use quite effectively. Of course, over time, I see the possibility of non-System i professionals and students looking for an inexpensive way of learning about the product firsthand as another group who would want this Little i on their desks.
Timothy Prickett Morgan, the editor of The Four Hundred and other IT Jungle newsletters, along with myself, would like to encourage the AS/400, iSeries, and System i community to weigh in with their thoughts on the ideas put forth. We expect that this article series will create a stir in the community and we are requesting input in the form of e-mails. We will also do a survey in the new year concerning the idea to get more input.
In my role of consultant and in my role as educator, the Little i can do wonders. I keep thinking of ways for the product to be more generally accepted both in academia and in business. Having a lot of them available to use is probably the best way.
I am in synch with TPM as he has written exhaustively about IBM lowering its prices to get more footprints, and in being more creative about how the System i5 platform is packaged. Price changes are not enough, as he has said many times.
I, too, think this is a very good idea. But, I would also like to see a more revolutionary undertaking–less cost, less risk, higher pay-off idea–and more importantly, an expedited time schedule. The notion of the System i workstation client is a sure success and if priced below $2,000 it would stand to become the dominant desktop (in a PC-like case) for application development on System i. It would be the System i developers’ tool of choice and so I would suggest that its initial packaging be for the System i developer, not the whole world. The whole world can wait until this guy does its thing for the “i” community. I see millions of these guys being sold, perhaps half that number being in the first year alone.
Does IBM have the power to do this?
Actually, IBM is the only power with power. Sorry, I just couldn’t help it. However, if IBM can convince Microsoft that it is OK to put three IBM Power chips in one little Xbox 360 game unit then Big Blue ought to be able to put the System i in a desktop PC case. If Apple could put a PowerPC in a desktop or deskside case, then IBM ought to be able to do it. IBM might argue that it can’t slip the Power chips into a teenie weenie white case and put it on your desk as easily as the guy down the street seems to make white boxes. But, look at the proof against this notion. The Xbox 360 is about 1/5 the size of a standard PC and the discontinued Power Mac G5 is already white cased size.
Wanting to find an existing IBM box that it could just tap and have the product already there (Voila!), I came across two nice IBM offerings. I was hoping to say, “See it almost already exists. Just make that existing Power boxes do it.” But, the standard IBM cases are a bit big and the top piece makes them a bit clumsy for the desktop. However, these IBM units are in the right price range for my 50 percent pervasive discount. In other words, if it is to be pervasive, reduce the price by 50 percent. Thus, the $4,000 p5 workstation in the below picture already checks in at $2,000 with the Kelly pervasive discount, of course.
The inimitable i5 with very little of anything to run anything with and a processor clocked down to 30/3800 (that’s 30 CPW out of 3800 available on the chip) of its potential power for interactive and 600/3800 for batch comes in at $12,000. A useful i5 configuration with the hardware capabilities of the AIX workstation below and the System p5 185 Express deskside model is really about $20,000 minimum.
We know that the i5 and the p5 are the same machine hardware-wise, but the position of “i” in the alphabet and the notion of integration together are the forces which compel IBM to charge substantially more for “i” than “p.” Sure, there might be other reasons. But, whatever they are, with the pervasive discount that I am willing to permit IBM to take credit for, even the AIX Open Power machine is $2,700 and that is in the ballpark for sure (though the results would be less pervasive than at $1,999).
The title of this article is “Fighting Pervasive with Pervasive: the Developer’s Dream Machine.” At $35,000, to even $20,000, to even $11,995, the System i is not priced to be pervasive. Therefore, it cannot be pervasive. Even the $11,995 unit is capable of handling substantially more than 20 users. IBM has no model for pricing a single unit System i intended for just one individual. How about $1,999 for 5 percent of the user capacity of the $11,995 model? How about $999 for 1/100 the power of the $20,000 unit? I have a few customers with the 30/600 CPW System i who have a base approaching 100 green-screen users. So, for 1/100 of the computing capability of this machine, how about a model for $999? It’s not too far fetched, but the $1,999 model would be fine for awhile. One secret ingredient for IBM is to get it out of the mammoth-sized black case that this little chip is locked into. Build a white case or use the Xbox 360 case plus a white case for I/O or use the old G5 cases, but get a new case. No one wants to put Stonehenge in the middle of their building, let alone their house.
Both $1,999 and $999 are pervasive marketing notions. In Marketing 101, it would be called a penetration strategy, if you will. This refers to a company realizing that the research and development is paid for and they can make the money on each box sold. In other words, that means no more skimming. A penetration strategy will pay off with a pervasive install base.
We’ll worry about the System i Developer Laptop another day, but for now, IBM, just get us into a desktop unit that cost about $1,999. I think that’s pretty clear. The OS/400 and i5/OS community will reward you with lots of orders.
Before I end this article I am going back one more time for the future, although the merits of the developer workstation should stand by themselves and signal IBM to commence the project. Plus, it is all I need right now though eventually I would make it support up to 5 users and this would carry the product into the mom-and-pop sized businesses. The Linux partition of course would take on the GUI. The office applications such as OpenOffice along with IBM’s Eclipse solutions under WDSc and iSeries Access client set this up for being a winning developers’ workstation.
The fact that it could then be afforded by me, by college students, by geeks of all flavors, and by anybody would help make it genuinely pervasive. To jump-start its presence in the universities, there could even be a special student package with free online education from IBM’s Academic Initiative. This incredible workstation would be affordable for the IT masses without making them forfeit their next car. Then System i mindshare (specifically, lack thereof) would not be the problem that it is today.
IBM can do it and easier than one might think.
Consider IBM’s initial PC experience. In 1980, Frank Cary, the company’s chairman, commissioned ten people to build IBM’s first PC from scratch. I would bet with a team of five or less, Rochester can come up with a desktop design of a Power-processor based Linux desktop with a built-in preconfigured System i and i5/OS partition in the background for application testing. I would buy one and so would most of my peers. Once available, its success would be assured and it would eventually morph into a Windows and Unix killer if IBM chose to permit that.
With a little box out there demonstrating affordable System i computing for the smallest entity, nobody would dare ask if the system were going away any day soon. Just having that question put to rest would give IBM executives more time to think of new and exciting uses for the System i. This new Little i has the potential to be one of the biggest Wow!-factor products ever to emerge from IBM and Rochester.
There is no better pervasive advertising notion than to do something so special that it makes the evening news. From announcements in the past, with Charley Chaplain and other famous well loved characters, if this one is done right it would bring in great publicity and positive publicity for the entire platform. The critical question would be “which i?” as the Big i and Little i compete for mindshare against each other.
If there were a way to eventually run XBox 360 or some comparable gaming software, it would make it an even bigger reason to buy from an emotional level, though the idea stands without this. Somebody would buy it just because it could do the Xbox. There’s always something appealing about a mixture of fun and business. This new Power PC System i Developer Workstation does sound a lot like the original PC. Now, I wonder what W.C. Fields would say if his likeness were chosen by IBM to be the eminent product spokesman for System i? Try these on:
If IBM chooses to create the dream System i, the resultant product workstation would not be a “PC” per se, but would be a model development machine that would not have the issues of Windows but would have has the full testing facility of IBM’s premiere GUI development environment built inside. Seamless partition integration would also go a long way in demonstrating IBM’s unmatched prowess in the virtualization area.
Is there any application developer who would not like the $2,000 developer dream machine? Can you imagine a factory-partitioned Power-based desktop developer’s workstation in which the partitions are so well integrated they escape the eye? That is virtually no work to set up is icing on the cake. Add factory WDSc and iSeries Access interfaces over virtual Ethernet connecting the two environments. Add all of the standard System i compilers, SQL, Query, etc., to run on the box enabling all source to be on the client side if you choose. Object code only would need to reside on the production System i. Wouldn’t that be the ideal development, testing, trialware, and training machine?
Since this is not the 1800s, you just add the Gnome or KDE variants of the X Windows environment to the invisible Linux. And, as part of this factory packaging, add the Open Office.org personal-ware along with a Linux GUI e-mail client and whatever else maybe necessary. The ideal world no longer needs to be installed with Windows. Would this Little i help the System i platform in any perceivable way? Can happy days be here again? At any rate, let us let the gurus decide. Do you want it? Send us an e-mail now, and take the survey in the new year when we ask you to.
Brian Kelly is an assistant professor in the Business Information Technology program at Marywood University, where he also serves as the System i technical advisor to the IT faculty. Kelly has developed and taught a number of college courses in the IT and business areas. He has an active consultancy in the information technology field. Kelly is the author of 27 books and he has written numerous magazine and e-magazine articles about current IT topics. He is a frequent speaker at the COMMON Users Group and he has been a featured speaker at numerous technical conferences and users group meetings across the United States.