IBM’s Reorg: The Good Me or the Bad Me?
August 13, 2007 Brian Kelly
IBM has revamped its marketing organization ostensibly “to better target the small-to-medium-sized business (SMB) market” as stated by corporate executives. The System i division had been part of the IBM Systems and Technology Group (STG). System p division had been part of the STG as well. Neither division exists in the new IBM. Instead, there is one sales group that sells large enterprise customers and another that sells to SMBs. As defined by IBM, SMB means businesses with no more than 1,000 employees.
Since IBM is in the services and software business (in addition to its systems or servers business), this reorganization is to permit other IBM product divisions to be able to sell their wares to the SMB customer set, many of whom are thought of as AS/400-type customers rather than full-service IBM customers. The change was made by Bill Zeitler, IBM’s senior vice president and group executive in charge of STG, who has been a driving force behind the recognition of the power of the System i line for many, many years. IBM noted that there will be a new level of sales management on the SMB side that will be created within each of 220 geographic sales territories in which IBM operates globally. The new SMB systems sales manager in each territory will focus solely on SMB sales efforts.
The two IBM players filling the big new jobs at the top are: Ross Mauri, who is to head a new Power Systems business unit within STG selling both high-end System p and System i products to enterprises and create Power-based systems and their AIX and i5/OS operating systems; and Marc Dupaquier, who is to run the new Business Systems unit and sell low-end System i and p products to SMBs, and is not responsible for developing anything. In the Power Systems business unit, IBM is introducing a new granular “pay for what you use” approach to pricing–or so it says. This is basically the mainframe approach to pricing, in which large customers have to decide ahead of time which features they get and which features they do not get to use. The SMB business unit will operate on an integrated system basis and offer full operating system, components, and hardware solutions on a per-user basis.
IBM’s public announcement noted that the company is making the changes to boost its market share in what industry analysts believe to be a $32 billion marketplace, based on 2006 sales. IBM sees its current share at 15 percent, so there is a lot of opportunity there if some solid programs stemming from this restructuring can help move the company forward. Clearly, it will be easier for a business partner that offers all of IBM’s products to operate in this new environment. In the past, for example, business partners had to go to five different IBM people in one territory. Under the reorganized structure, they will just have one.
The Bad Me Has a Bad Opinion of IBM’s Recent Moves
Since I always like to hear the bad news first, let me get it out of the way.
IBM’s recent reorganization is clearly a recognition by the company that trying to go back to the glory days of the System/38 was not working. In fact, it is not working even a little bit. You may recall that in the 1980s, IBM wiped out all of the minicomputer vendors in the world with its AS/400 product launch. Where did that IBM go? The competition did not know what hit them back then. Soon, however, IBM stopped paying attention to the AS/400 while it was trying to exploit its PS/2 and OS/2 one-two product punch for the client desktop. To those working the sidelines in those days, the one-two punch turned out to be much less awesome than the Giuliani Girls vs. the Obama Girls recent pillow fight. Microsoft, Intel, Compaq, and others knew IBM had guns loaded with blanks with its marketing approach for PS/2 machines and the OS/2 operating system, and they went on to reduce IBM’s PC market share to almost nothing and eventually forced IBM out of the desktop PC business.
IBM stumbled big-time in the 1990s. John Akers almost sold the corporation to the Grim Reaper piece by piece, losing $13 billion in his last two years. Many of us believe that if it were not for Lou Gerstner’s focus on services and software, HP would have been able to take over IBM in 1995 for chump change. Lost in the shuffle was this fabulous AS/400 box that was still knocking them dead in what was known as the “midrange” market.
The PC was not the only battle IBM was losing during this period. The venerable SNA/SDLC communication architecture and protocol was never really adopted by the government or academia, though it had been wildly successful in the business community since 1974. The AS/400 as a “new box” was all SNA and was often the first with a powerful new SNA facility such as Advanced Peer to Peer Networking. IBM had a TCP/IP “product” available for the system that was terrible, basically a $26,000 feature that permitted IBM to bid on government and academic projects. The TCP/IP was so bad that if it were ever essential for the success of the installation, IBM would not bid–and it cost $26,000 with no tiered pricing.
Unfortunately, IBM had made the AS/400 an SNA-only machine at its heart and even with the TCP/IP add-on, most of what one would expect in a viable TCP/IP stack was not to be found. And, as you may recall, the Internet was coming into its own and then-Vice President Al Gore was pushing the notion of the “Information Superhighway.” Of course, the highway was paved with TCP/IP, so the AS/400 was a non-participant in the early 1990s.
Additionally, the client/server revolution was upon us. Unix and NetWare servers brought forth file and printer sharing for a PC-crazed neophyte business user class. Additionally, Visual Basic and other visual languages permitted users to benefit from the same point-and-click interface they had come to enjoy with Windows 3.0, 3.1, and then Windows 95. To choose not to satisfy these needs could indeed be the death of any platform.
For both the Internet and client server, the IBM corporate strategy was to keep the AS/400 as a non-player in this critically important application area. IBM’s rationale was simple. The company already made two platforms–the RS/6000 running Unix, and PC servers running Windows for Workgroups, then Windows NT, as well as NetWare–so its executives believed that two out of its four lines were enough to satisfy its customer’s needs. IBM has been paying the price in lost sales to this day.
You and I would not have made such a decision. It was like a GM executive deciding that the Oldsmobile, Buick, and Cadillac lines did not need automatic transmissions since the Chevy and Pontiac lines offered this modern convenience. Why should the premiere divisions retool for this convenience when the less expensive models already had this stuff built-in? You and I can see that those who would have ordered Cadillacs, of course, would not have bought a Pontiac as an alternative, but instead would have opted for a nice Lincoln Town Car or a Mercedes. GM would register this as the potential for lost sales and thus would never offer important features only on their lower cost products.
You know that. I know that. And Lou Gerstner knew that even before he was briefed. Lou Gerstner, who was a first-class merchant made a decision shortly after his arrival at IBM on April Fool’s Day, 1993, that would help stop the AS/400 and mainframe product line erosion (remember the mainframe was dead for awhile). Yes, IBM had also decided that its mainframe division did not need a robust TCP/IP stack or client server facilities. But, a guy who had just spent a part of a lifetime working with cookies and cigarettes somehow saw something that the cream of the crop IBM executive team did not see or would not see.
Gerstner immediately called for the retooling of all IBM systems to fully support the Internet and client/server. He not only saved the corporation with his services and software strategy, he personally saved the AS/400 and mainframe divisions by forcing the line executives to add the missing (and unbudgeted) TCP/IP stack and client/server facilities to the product lines.
So, in 1994 with CISC V3R1 and V3R2, IBM introduced a new, no-charge TCP/IP stack to the AS/400. It performed 700 percent (eight times) better than the former chargeable stack. The same team that built the stack for the AS/400 (operating from Endicott, New York, under the able direction of Armando Fratezi), built the same stack for the mainframe. At the same time, IBM introduced Client Access/400 and the integrated Intel server to help its AS/400 customers with the client/server revolution that by then had just about completely passed by. (You may recall it was called the File Serving IOP way back then.)
It took until the late 1990s for the full stack and the well-known TCP/IP utilities to be available with expected function. The 1994 offerings were quite buggy as IBMers new to the TCP/IP protocol were implementing to spec and not to how the Unix players in the industry had actually shaped the protocols to perform. Nothing happens overnight and thus, while the dot-com revolution was in full swing and companies were spending huge amounts for Internet servers, both the AS/400 and the mainframe were just being made capable. Can you imagine the lost sales to Sun Microsystems and others who championed the dot-com day? Yet, other than John Akers himself, no IBM manager I am aware of was released from duty.
Meanwhile, AS/400 developers and development teams were disenfranchised from participating in the dot-com revolution. It eventually became a dot-com bust for sure, but there were some great days and there are some great companies that got their roots during this period. Just look at Amazon, Yahoo, and Google. These companies were not able to enjoy the benefits of an AS/400 at the time, since the AS/400 was still incapable of meeting their needs.
Finally by the year 2000, most of the TCP/IP artifacts necessary for a real Internet server found their way to the AS/400 product line. Most of the new development work on the system was still green screen, despite the fact that the system could now support TCP/IP and all its wonders. Why was this? With Gerstner no longer paying attention and perhaps believing that all systems were enfranchised for the Internet and client/server, he issued no more decrees. Yet, the last step of the metamorphosis had not taken place–the ability for AS/400 developers to develop AS/400 style and have their end-products usable on the Web.
Why did this happen? Corporate IBM began a love affair with Java that continues to this day in one form or another. Somebody actually thought it was OK to tell the AS/400 developers to learn German when they spoke only Spanish. Somebody in IBM thought, and apparently still thinks, this is OK.
AS/400 applications are green screen and they still are developed using the innovative Model View Controller (MVC) notion that has been prevalent in AS/400 since the System/38. The integrated database provides the model and the innovative display file object (always separate from the mainline program) provides the view, and RPG or COBOL statements provide the controller (logic). For some reason, after IBM enabled client/server on the AS/400, it chose not to take the last step and update the “V” part of MVC. As we speak, the “V” in MVC is still green screen. Seven years after Lou Gerstner had to command IBM to bring the Internet to the AS/400, they still did not get it and still, in 2000 had no natural view for the system to augment the notion of the display file. Seven years after 2000, in 2007, this facility is yet to be announced. What will the next seven years bring?
Instead of updating its operating system and its display file notion with real Web artifacts, IBM fell in love with Java. This time, the IBM benefactors dictated that the AS/400 must be a Java machine. The irony is that by including the AS/400 as part of the Java revolution, IBM actually disenfranchised the bulk of its AS/400 developer community from the Web. Its preoccupation with Java for so long kept the resources from building the display file replacement–the Web file. The Web file still is not here.
IBM believed that it had solved the Web problem on the AS/400. Hey, if Java works for the Unix nerds and Windows weenies, why not the AS/400 crowd? With all IBM systems speaking Java, a virtualized C++ language developed by Sun Microsystems in 1995, IBM was then in a position to support all systems by supporting one language. “So let it be written, so let it be law.” Though this strategy was not working, undaunted, IBM went off to more important projects such as homogenization and renaming.
In 2000, we saw the new eServer iSeries 400 re-introduced after it had actually been introduced earlier in the year as the AS/400 8XX series. That brought with it the controversy of whether an 8XX machine was actually an iSeries or was it really a renamed AS/400 or was it neither. This was not IBM’s finest hour.
The company had been getting beaten up by the trade press for having four disparate product lines–mainframe, Unix, Windows, and AS/400. So, by calling each and every one of them an eServer, IBM solved the problem. There was just one problem remaining: The systems were still disparate simply operating under the same family name. Some believed that “IBM” was the family name and did not understand the need for the eServer moniker. It was kind of redundant. Was eServer really a middle name? Many such conundrums ensued and IBM looked sillier and sillier. As an analogy, the GM Cadillac was renamed as the GM Automobile iSeries 400. The GM Pontiac became the GM Automobile xSeries (no number). Only long-time followers of the product line were actually interested in what IBM had done and only they understood what was what.
Nobody could explain what an AS/400 was except by going back to the old name. The mainframe became the zSeries. The X86-based servers became the xSeries, which at least made sense, sort of. The Unix box became the pSeries, the p being for “power” and “performance.” And what was it that had become the iSeries? IBM could not nor world not answer that its finest business system was now the iSeries. So, it tried to say that its integrated (i) system was the iSeries. But what did that really mean? New account sales just about died during this period as the systems and its user community searched for a real place and a real identity within the product line.
The handwriting had been on the wall for some time and AS/400 stalwarts such as Al Barsa and Bob Tipton had been screaming at COMMON meetings for IBM to market the AS/400 to increase market share. Even before the eServer announcement, in fact, the troops were upset at IBM’s performance in marketing the AS/400 system. Let’s take a peek at one of IBM’s ads that helped spur a negative reaction in the AS/400 community. This is a print ad that Big Blue was using in its June 1999 ad campaign:
“What if there was a box . . . a magic box? A box that contained all the answers to all the questions you’ve ever had . . . . There is such a box . . . . A magic box. It’s called a server. . . .”
I can tell you this: My neighbors were not running over to my house that day asking what that ad was all about. Where was the part about the AS/400? The ad, when seen by the AS/400 community, read differently:
“What if there was a box, a tragic box, a box in which you could bury your product’s uniqueness among a sea of conformity? There is such a box. It’s called a coffin. And this campaign is driving a nail in the lid.”
You may recall these as the words of Neil Palmer, an AS/400 technical specialist at DPS Canada Limited, as repeated in the halls after a COMMON session in San Antonio, Texas. The eServer campaign brought about the full marketing homogenization of the product lines and the AS/400 got lost. The renaming was a bust and iSeries sales from that point on were a bust and there are those who believe the AS/400 (now System i) will never recover from this IBM action.
Java has since failed IBM in its desire to bring the Web to the AS/400 community. AS/400 developers rejected it. It’s that simple. IBM still doesn’t get it. But IBMers do not blame themselves for the decline of the platform. They think perhaps that just maybe they didn’t explain it well enough for the type of developer in the AS/400 community or perhaps most AS/400 customers are too thick to know that IBM knows best–or both. Actually, the truth is that AS/400 developers are so busy doing their jobs (developing) in the most productive language of all time (RPG) that they had and still have no time to downgrade their shop’s capabilities to the lower-level language of Java.
So, when was the AS/400, iSeries, i5, or System i’s last success? I don’t really know. The hardware is great, but then again, so is System p hardware. RPG, the language of this community, has not been enhanced and rumor is that IBM will be asking its System i developers to have faith again to give the company time to bring out a new IBM product that is better than RPG. So there won’t be any more RPG upgrades and that Web RPG I have been dreaming about; that ain’t coming either. There is a product in the works called EGL that “has everything that you want in RPG,” plus, and here is the kicker, just like Java, it runs on all platforms. Sing this song, “Where have all the systems gone? When will they ever learn?
So, the Bad Me recounts all of this bad history and feels compelled to conclude that this is a way to let both the big and small side of the System i slip into oblivion. I can see the old eServer moniker coming back as the machines on both the big and small side of the aisle are called eServers and this time, they actually are all the same. Soon with Power6, or Power7, or even the mainframe will be on the large side of the aisle.
It looks like there are just two things that differentiate these boxes now: marketing and the operating system. With a green-screen OS as the only face of the System i, four name changes later and more than 15 years after Microsoft brought out usable GUI and took out Novell with it and chicanery, IBM still doesn’t get it. IBM will be winning no user interface awards for awhile as legacy is what it will continue to build and legacy continues to be a bad word.
Back to the car analogy for a minute. Would GM ever choose to sell its Escalade without a paint job or a quality set of tires? No matter what is under the hood, an ugly first impression does not make one want to look any further. And, as for IBM marketing, there is just one other thing uglier than the look and feel of the operating system.
With no marketing and no pizzazz, the new organization permits the System i and all of its prior incarnations to disappear quietly and nobody in IBM will get punished. The company that will miss the system the most is probably Microsoft, which finally will be forced to figure out how to run its business without a few dozen AS/400s hidden in the back room.
The Bad Me is very vocal, but you’ve got to admit there is a lot of ammunition available. I sure hope the Bad Me is all wrong on this.
The Good Me Is Looking for Big Success!
The Good Me looks at this reorganization as IBM has sold it to the press and feels some hope. Big customers and small customers do have different needs and this announcement helps them all.
On the big side, the Virtualization Engine hypervisor, which had been extracted years ago from the mainframe’s VM hypervisor and buried in OS/400 microcode and then put even deeper underneath both i5/OS and AIX–is the magic piece that permits all of the hardware system homogenization to prosper while promoting an OS-neutral environment. Even the standalone System i 515 and 525 boxes of today operate with a hypervisor as the first OS layer (or the last hardware layer, depending on your perspective). IBM will use its relationship with VMware to put out a hypervisor that can talk to storage on the System i while running on System x and BladeCenter blade servers, permitting the old Fort Knox dream to come true in a sense. Systems sold through the Business Systems and Power Systems divisions will benefit from these and other changes.
The Business Systems division should be able to put some marketing programs together to leverage this unique capability. I would expect that IBM will actually permit this new group to advertise its wares rather than have customers guess what system they should buy. The Power Systems division will be able to sell cooperative and harmonious processing on all operating systems to all levels of granularity. The cluster is the system. It is a grand day for IBM hardware.
The Business Systems group will look around and say, “Hey, what can we do to whiz-bang this system and give it some pizzazz so that it does not need so much explanation to sell?” One technique, of course, would be to add a GUI to the hypervisor. Put the view component for all OS functions in an expanded hypervisor. Some of you may recall that VM actually became an OS unto itself because it made sense sometimes to run CICS or DB2 database right out of the hypervisor– whoops–Virtual Machine operating system on a mainframe.
Now, once this has happened, it means that new applications from any system can be accommodated via the GUI OS links in the hypervisor. If 5250 is moved to the hypervisor, it also means that any OS would be able to drive 5250 applications. Hey, who cares about those?
In truth, the bottom line is that’s all we care about. Aren’t the applications that run all businesses the reason that owners, presidents, chief financial officers, and other corporate bean counters part with gobs of cash? Wouldn’t it be good to have 5250 capabilities on all platforms using these hypervisor facilities?
It would also be nice to have an RPG and CL compiler for each platform. If the hypervisor provides OS-neutral 5250 and if the hypervisor provides OS-neutral DB2 facilities, created via DDS or SQL, then it ought to be fairly simple to create a command language that supports CL for Linux and AIX, and perhaps even Windows and, OK, how about x/OS?
Only if, as a developer, you feel you need the capability of single-level storage memory addressing, a technology independent machine interface (whoops–isn’t that the hypervisor?) or some other unique System i facility, would you really care which operating system is running your applications. And, quite frankly, I would give up all of the OS-specific whiz bang integrated computer science notions championed by Frank Soltis to be able to have a business language like a new Web-based, natural, graphical RPG run on all IBM supported operating systems. I would take that rather than just one OS with single-level storage–as much as I like single level store–because single-level storage will not sell one extra system or processor, but a Web-based RPG on all platforms will bring in big bucks from multiple sources. There can be development on all platforms that is so rapid that it will blow the old way absolutely away.
And, while I am being greedy in my notion of what a successful STG division should look like, I would suggest that the RPG language name be changed to The Business Language. This reflects the real purpose and meaning of the language. I wouldn’t be unhappy with a name such as All Business Language, either. This multi-platform natural Web GUI language deserves more than being dubbed the next-in-line son of RPG.
I don’t even care whether the Business Systems division or Power Systems division sells the box. With these capabilities, sales opportunities will follow and maybe–just maybe–the IBM cash registers will start posting big-time transactions again.