The Web Runtime Tax: The Tax Man Cometh, Again
June 18, 2007 Brian Kelly
On February 5, I had the pleasure of applying for a U.S. passport for an exotic cruise I am taking this summer, but I was hoping that I might expedite the process so that I could use the passport to fly into Toronto, Canada, for a one-day IBM pre-announcement briefing on iSeries Application Development-the WebSphere Development Studio Client (WDSc) in particular.
Unfortunately for me, without a trip to Philadelphia or Harrisburg, the time for which I could ill afford, my passport would not arrive in time to fly to the February 22, 2007, event. In mid-January, the rules changed and flights into Canada require a passport but you can still drive across the border with photo ID and a birth certificate until 2008. So, my 26-year-old son, who recently passed the Pennsylvania bar exam, accompanied me on the drive to Toronto, and we enjoyed some quality time on the trip. I emailed IBM‘s Kou-Mei Lui, the Software Group’s marketing manager for WebSphere Development Studio, WDSc and Advanced Edition, and WDHT, that my son was accompanying me on the drive and she graciously added him to my company profile so he would be able to enjoy this significant pre-announcement session as a disclosed participant.
Along with about 15 others from around the globe, I had the pleasure of being invited to the RPG and application development capitol of the world for this special briefing on the upcoming product announcements and technology direction for the System i platform. IBM’s leaders Kou-Mei Lui, George Farr (solution manager of WDS, WDSc, and the RPG compiler for the System i), hosted the event. The speakers were from George’s team. They were the IBM experts in the many areas of System i AD. The session was well done. The speakers were very knowledgeable and I left encouraged but concerned.
IBM displayed the many faces of WDSc V7 along with WDHT at this session and previewed some exciting (and long overdue) things about the RPG language that I have been waiting a long time for. With IBM releases on a two-year cycle, it looks like it will be another year before the real application development revolution begins, and like many who follow IBM’s moves in the System i space, I am doing my part to help assure that when it happens it is not a big dud like so many mostly wonderful but incomplete AD offerings from our friends in Software Group.
“Mostly wonderful” is how I think I might describe my reaction to the V7 WDSc bundles that were formally announced February 27 and are now available for downloading or ordering with help from your System i Business Partner. The cup is mostly full. However, the fact that there is any degree of emptiness seven years after IBM announced the WebSphere Development Tools (WDT) is cause for deep concern. Historians know the WDT was quickly superseded by the WebSphere Development Studio Client (WDSc) packaging and this name has survived. Though the name is the same, with Eclipse and Rational now providing the structure and the engine, the underpinnings and the look and feel of this product package have changed substantially over these many years.
Having my own perspective on the matter from the Toronto session, I read with great interest Bruce Guetzkow’s excellent article in this publication (“WDSc Version 7.0 Standard Edition Is Missing Two Key Features”) in which he looked at the WDSc V7 offerings point by point. In reading the article, you could get the sense that there was a little tear in Bruce’s eye as he looked out and like the Native American in the litter ads years ago, he could not believe that anybody would do this to anybody else. This newsletter ran feedback from readers on Guetzkow’s article. (See “System i Shops Plenty Annoyed About Missing WDSc Features”) Customers were bewildered that, even before IBM sold the System i community on using its almost complete toolset, it was stripping needed function from the offering. Moreover, along with the obviously missing pieces, Big Blue also placed other very desirable facilities into a separate, high-priced product dubbed the Advanced Edition. Who are these features for? Despite the obvious heat in the reader emails, Guetzkow answered each and every note that was published with a very thoughtful response. Hopefully, the eyes and ears of IBM are more than watching and listening because, in my opinion, IBM cannot afford another false start in this strategic area. IBM has been hinting in the wake of this turmoil that these features will be priced separately with a nominal fee for Standard Edition users, but so far as we know, nothing has happened yet.
I wish that from my Toronto experience I could offer assurances that IBM is acting. IBM saying that it is going to address the missing GUI designer is really just half of the issue. The other half of the problem is perhaps even bigger. In Toronto, for example, even before the products set was announced, there was grumbling from the briefing audience as well as those who were conferenced-in from their home locations as the Advanced Edition was highlighted. However, this was not a wholly interactive session. IBM was looking for some input but not about its new cost structure. Big Blue’s purposely brought us to the development lab instead of the marketing office. It wanted for us to learn the new facilities that are in the release and to show us how close IBM Toronto is to having this marvelous seven-year work in process completed. Now is a good time for that marketing critique.
I cannot ameliorate some of the negative feelings that exist about the PDM/SEU/SDA replacement being incomplete in the standard edition (the edition for the masses). I don’t understand it from a System i perspective and in a word, I think it stinks. My purpose for writing this article, however, is to point out there is a lot more wrong than two missing pieces. There is a big hole in what appears to be the whole strategy for modern AD with the System i. And, yes, as I will show you later in this article, history does repeat itself. These same mistakes were made before and the last set of mistakes derailed the Web for iSeries clients by as much as three years. I do not in any way want to minimize the two missing pieces that Guetzkow pointed out in his article, but the fact is that SDA can still be used instead of WDSc–so at least there is a tool that can and is being used.
Hidden in its announcement letter 207-030, yet in full view under a caption that includes the word, benefits, IBM boldly states the following bad news in bullet form:
“Uses IBM WebFacing Deployment Tool for WebSphere Development Studio with HATS technology V7.0 run-time license for deploying WebFacing, HATS, and linked WebFacing/HATS applications to production.”
I underlined the words “run-time license” because this is the essence of the newest bad news. On top of the insult and some big injury, there is a big hurt. As this product set approaches being fully built, IBM has decided to tax customers who choose to use it. There is no tax, however, if you don’t use it and so I fear that is the road that many System i shops will take, thus again taking the Web wagon off its rails. As the community is digesting what some might call “price gouging” with the tax on Web runtime, there is a pall in the System i community with much disappointment, resentment, and in some cases anger.
As I was reading intently, I learned that this “software product set” is now marketed under an IBM software license called Passport Advantage. Ironically, I am finding about as much advantage to this as my own personal passport experience in getting to Canada for the announcement briefing. Passport Advantage to me is just a euphemism. I am convinced that it merely means that the IBM Software Group is in charge of these offerings–not the System i division. Passport Advantage is how they do business.
Though IBM’s announcement letters have been corrected post facto with the V7 product codes, I still do not know and have no way to find out from my usual IBM sources (who shut me down because I ask too many questions) whether the same runtime charges that were announced for HATS alone last year applies to V7 WDSc/WDHT. But, it certainly seems so. And, my business partner is convinced that it is. So, here are the tiered prices:
WebSphere WebFacing Deployment Tool
In IBM’s Quick Pricer, 5722-WDS lists for $3,650 for tier P05. The WDHT component costs $5,000 at software tier P05 (the smallest tier). Thus, for only $8,650 extra, the new System i 515 with five users at $7,995 can be equipped with the development environment for WDSc and HATS and it can have the runtime environment–all for “just” $8,650, thereby making the $7,995 entry price $16,645. The combination of the compiler and tools package and the i5 515 doubles its cost. One would expect that for a new account hoping to have a few applications running in Web GUI mode, the compiler package would be necessary and the runtime environment. It seems that this can take the wind out of a lot of potential new account System i sales and thus is not something that makes me happy.
There is also a per-user per year option of $120, which can be selected instead of the tiered price. For the i5 515, this would be the better option at five users, but gets expensive over six years as the cost is $720 per user. Considering that software versions last quite a while (we’ve been on V5 for about six years now), the total per-user charge for six years of $720 gives a breakeven at tier P05 between seven and eight users–not considering the time value of money. That’s not chump change.
Last spring, IT Jungle’s editor in chief, Timothy Prickett Morgan, offered his commentary on phase one of the transition to a charge basis for the once free HATS (provided originally as HATS/LE, or Limited Edition). (See “IBM Weaves Together HATS and WebFacing Tools”.) It looked like IBM was going to let WebFacing stay free at the time, as Prickett Morgan opined, but in this article he paraphrased IBM’s warning. “You do not need to suddenly buy WDHT if you plan to just use WebFacing, by the way, but IBM warns that in future releases of the WebFacing Tool, all deployments will require the purchase of WDHT.”
To the tune of McCartney’s “Yesterday,” under my breath I can’t help singing, “Oh suddenly came suddenly!”
OK, since you liked that, here is some more:
“Why IBM had to charge for runtime
Suddenly did come suddenly–just a year later. It came at a time when the System i marketers had figured out how to build a runtime option that would drive both HATS ad-hoc and WebFaced pre-generated applications together with no interactive penalty. What that means is that if the only reason you need interactive is for HATS-driven Web applications, then you would not need to buy the Enterprise Edition of the System i that, because it provides no perceived value to many, has been called the Interactive Software Tax (a phrase coined by TPM in 1993, by the way, when IBM first started differentiating between AS/400 “systems” with interactive processing and AS/400 “servers” without it). Of course with the i5 515 and 525 machines, the Enterprise edition is so passé.
To put this interactive tax in perspective: At the P10 level, the cost for 1,000 CPWs of interactive power is $36,000 on 520 and at the P20 level for 3800 CPW of power it costs $90,000 per processor. The 515 and 525 change this metric for the better but many System i small shops have already paid the interactive tax. For that premium, on IBM’s formerly smallest systems (the 520), the Enterprise Edition feature was often twice the cost of the rest of the system. It sells only because of fear. Nobody wants to get caught short on interactive power. So, many companies are ponying up for the Enterprise Edition rather than taking the risk of not having enough. Having an interactive slowdown that you caused during a critical run just because you didn’t want your company to pay the tax would not bring about a desirable day for you if you are the IT Director who made that decision.
The reason companies pay the interactive tax is that only once paid will IBM permit the Power 5/5+ chip that you thought you purchased free and clear with the machine to be used for interactive work at rated speed. Apparently it was IBM’s way of extracting additional revenue from its customers for using IBM’s proprietary interfaces rather than switching to Windows interfaces or “modern interfaces” as IBM calls them. I am glad this tax is gone with the 515 and 525. The irony with this notion always was that if you do not move your company to the Windows methods-batch, client server, or Web technologies from 5250–you had to pay the interactive the tax.
The IBM pricing experts set the price so high for full CPU power ($90,000) that in order to sell the machine to smaller customers who need interactive but cannot afford $90,000, the company had to create a special deal. This deal downgraded the speed of the whole Power chip to about one-quarter speed (1,000/3,800 CPWs) and it provided this downgraded chip. So, you get 1,000 CPWs in batch, client/server, and the Web, and you also get 1,000 CPWs for interactive in what I like to call this “semi-enterprise” scheme.
This less expensive, semi-enterprise edition costs $36,000. If you check the proportions, the small user shop got stung on this more than the big guys. For the full chip (100 percent) in batch and interactive the cost was $90,000. You get 100 percent of the chip for 100 percent of the charge. If you didn’t have $90,000 handy, then you had to settle for 1000 CPW which is about 26 percent of the chip’s power, the semi-enterprise edition rang up at $36,000 more or 40 percent of the cost. Smaller customers got 26 percent of something after they paid 40 percent for it. There was no way to look at this other than that fast interactive processing is expensive on System i–by design.
Are the Power5+ chips that expensive? No, they are not. I would expect that the IBM Microelectronics unit makes some profit also, since IBM is not a charity. So, how much profit is there in selling 26 percent of one-half a chip (one processor core) for $36,000 and 100 percent of one-half a chip for $90,000? Of course if you have a two processor engine, with the interactive tax applied to both processors, the question is how much profit is there in selling 100 percent of a whole chip for $180,000? Adding a bit of an insult to the loyal System i customer set, you do not pay this tax if you choose to run your System i as a Unix or Windows box.
Of course customers never really had to buy the Enterprise Edition, you could have bought a Standard, Value, or Express edition. All of these were schemes to either remove interactive processing from a model (standard edition) or make interactive more affordable in smaller chunks for smaller companies. Even on the Value and Express editions, there is an interactive tax but since IBM provides so little of the chip for interactive, it is not close to $180,000.
I have a client who just moved from an Advanced/36 to a 30/500 CPW 520 Express Edition. It cost about $22,000 including software–with a few extra disk drives and some more memory than standard. This customer has about 100 users with System i 5250 interactive access, who are not data entry people but they do daily work. This system is orders of magnitude faster than their Advanced/36 model, which they just moved into the parking lot.
I have another account that needed bigger disks and even more memory plus they needed the full CPW of the chip to perform batch and backup activities with a minimal amount of downtime. This value edition client paid about $100,000 for the System i 520. In this total is something called the Turbo edition (a.k.a. a turbo tax–not usable for tax returns or interactive–just for batch, client/server, and Web workloads). This enabled their batch, client/server, and Web CPW to be at full one-half Power5+ chip capacity (3,800 CPW).
This customer’s machine had 60 CPWs of interactive capacity version, and they paid a larger interactive tax for the privilege. At 60 CPW, they got 60 divided by 3800 or just 1.5 percent of a CPU core. My earlier client, with 30 CPWs, supports its 100 users on 0.75 percent of the Power5+ chip. That’s how fast the processors are and that’s really how many green screen users you can run on these small systems.
In all of these interactive tax schemes, you pay more for interactive than if you run your machine in a Web or Windows style. With the turbo tax the irony is you pay more for batch, so maybe this turbo tax is a really a batch tax. Now, there is a new tax to watch out for that is easier to avoid if you’ve already paid your taxes. I call it the Web runtime tax, and it requires you to pay more for running your applications in a Windows/Unix style on the Web. You pay it on the older 520 machines even if you already paid the big interactive tax and you pay it on the new 515 and the 525 machines that are not tax interactive. I know these interactive taxes were levied originally to make System i heritage shops move to the Web technologies that were prevalent in Unix and Windows systems, but why a penalty now for going to the Web?
So, why a Web runtime tax? The simple answer is because it is a tax and IBM has the authority to levy such taxes. For anybody who thinks they bought the whole chip with their system, think again. For anybody who thinks they really got a full set of AD tools with 5722-WDS, think again. The tax man cometh. Even if you take IBM’s advice and get off interactive and go to the Web, there is a special tariff designed just for you. IBM’s pricing experts have become experts in taxation.
A number of industry analysts have noted publicly or privately that since IBM no longer charges HATS transactions against the interactive CPW bogey, the Web runtime tax really is free since you don’t have to pay the significantly larger interactive tax. However, that logic is flawed. You either already own your System i or you are buying a 515 or a 525 that has no interactive tax. You are not reading this article because you are a Windows shop thinking about getting a System i.
I would suspect that like me, many of you have your own System i. It’s already bought and paid for. Therefore, any tax you pay now is extra. There is no savings. When you bought the system, HATS/LE was free and WebFacing was free and there was no Web runtime tax. You could develop your applications and then you could run them for “free.” Now, to run either of these, you have to fork over as much as $18,000 per processor for the same rights as you had before February 27.
IBM is not refunding any of the interactive tax money that it collected. Moreover, if you bought the Enterprise Edition just so you could run HATS, you already spent $36,000 to $90,000 more than you probably need. You aren’t getting that back. Now, to be able to use this investment, you must get your wallet out again to pay the Web runtime tax.
My small 520 Express Edition client with 30/500 CPW may be too small a system to run WebSphere right now, so this customer probably won’t be paying anything to IBM in the short-term to keep running their batch and be interactive. However, I have been suggesting that they begin to try a few WebFaced applications over the Internet and see how it goes. A month ago, the customer could take a look at the performance impact and see if that is something it should continue or not. Now this customer finds out it cost them $5,000 of unbudgeted funds, or they can try the two free users and get their clients upset when user three gets rejected.
This customer paid just $22,000 for the whole system, and it if needs more power to do more things, it does have the turbo tax option of the Express Edition. This costs about $13,000 to get almost a full chip’s worth of power for batch and the Web–up to 3100 CPW on the 30/500 i5 520 upgrade charts. Well, not exactly! If they choose to do that, the Web runtime tax now becomes $10,000 instead of $5,000 because the Turbo tax also puts the system software at level P10. It seems to me that if the objective is to be more Windows like, this extra tax plus the extra fee for a P05 to P10 software upgrade may be just enough to convince a small account that .NET running on a Windows machine, rather than a Windows-like machine, has a lot of financial advantages. Hey, it’s the real thing.
For my larger Value Edition 60/3800 CPW customer, the idea that HATS would run in the batch factor of CPW (3,800) would permit the company to have the power of the whole chip to run WebFacing and HATS combination applications. That would be really nice, but now this customer has to go back to the well for the tax money since IBM added a new levy. Though not as drastic a percentage increase as the small guy who really gets killed by the many taxes, the WDSc V7 announcement means that even if the larger small client chooses to stick just with WebFacing and forego the HATS panel to page on-the-fly translation, they still must find another $10,000 from a budget that just provided $100,000 to buy a nice sized model 520. That is a lot of money for a shop that must come up with the $100,000 every five years at upgrade time. And it’s bad news for me since soon they will be asking this friendly consultant why I hadn’t included that in the price of the system that they installed at the end of last year. I am not looking forward to that round of discussion. This is a lose-lose proposition. Even IBM is not a winner here, because this is the sort of thing that agitates customers.
As bad as that is to my client set, the most unfair hit, though they may be able to afford it more than my small clients, is the Web runtime tax per processor for larger systems. At 3,800 CPW, a two-core i5 520–even if the one core is not activated–is at the P20 software tier. Is a 3,800 CPW processor not a 3,800 CPW processor? At P20, the Web runtime tax is $12,000. And, if the system happens to be bigger in size with even more of these 3,800 CPW processors on the model 550s, 570s or 595s, for the same CPW, these shops pay $16,000 per processor in Web runtime tax.
Is this the kind of publicity IBM was interested in when it made this announcement? I don’t think so. Hey, I like the technical guys in Toronto and the marketing people are some of the nicest human beings I have ever met. They are first-class people. They have a job to do. I noticed the sensitivity regarding the runtime charge in the room–from IBM and from the customers and clients. I saw faces wince at the thought that IBM gave back CPW cycles with HATS running on the batch side and then took the savings away by the runtime charge. I saw a puzzlement that could not be solved at a Software Group level. What’s good for one division is not necessarily good for another division.
Quite frankly, I am tired of the Software Group ruining System i parties. About 10 years ago, when the Software Group was formed, it was jump-started by then chairman and chief executive officer, Louis Gerstner. Software Group got to take over any software product that was not an operating system. How convenient that a nice, productive, and profitable software lab in Toronto, separated by many miles and a country border from Rochester, Minnesota, was already in place. It was a quick switch to move it to the Software Group. It was a bad idea then and it is a bad idea now. It’s time Rochester took it back. Unfortunately, Software Group is big enough to buy and sell the System i division, so that logic has a few flaws in it. But, then again, there is a new chairman and CEO at IBM, isn’t there? Sounds like this is a job made for Sam Palmisano.
When I worked at IBM many years ago, and I delivered the product excellence presentations for the System/38 and then the AS/400 new account prospects, I had my own definition for “integrated.” Knowing that every product in their shop was a silo just like the mainframe, I would offer that: “if it has a name, the developers didn’t know about it when they built the system.” Therefore, it is not integrated; it is made of piece parts. Software Group makes its money on piece parts. A piece parts strategy is incompatible with the System i notion of integration. Both the Web runtime tax and WDSc Advanced Edition are piece parts ideas that just don’t fit in an integrated division.
The 5722-WDS product came into being six years ago with an implied promise that IBM would better assure that no longer would there be haves and have-nots in System i application development. Other than SQL, all the development tools moved to WDS and it was a brilliant move. It is easy to see the harm that SQL as a separate product did for System i shops. Even today, many customers do not have it loaded on their systems. Thus, System i is the only relational database system in the universe in which most developers do not or cannot use SQL. That is another big cost of a piece parts strategy–lack of uniformity. System i shops don’t want that. We want integration.
Many shops are accustomed to paying software maintenance for their products. SAP and other application vendors collect a hefty fee for this maintenance. IBM always said that it used forward pricing in that it charged today for enhancements that it was going to build tomorrow. Tomorrow is here and now there is a Web runtime tax. There is also a software maintenance charge. Apparently the charge is not substantial enough to fund the new development work and make a nice profit for the Software Group. So, figure out how to sell more systems. That would help and, if IBM must, make the software maintenance charge a bit bigger but don’t price the system out of the game. Do anything but change the System i into a piece parts system.
Over a five-year period, from 2001, I wrote seven books specifically on how to use the many versions of IBM’s WebSphere tools. Unfortunately for me and for IBM, by the time each book came out (and I write pretty fast) most of the material was outdated by either a new version, release or modification level. In fact, I had one book that was out of date overnight–a result of applying a new function PTF. I actually witnessed the look and feel change before my eyes–so much so that it no longer matched the screen captures in my book. Because of the volatility of the product set I no longer write WebSphere books. All of my books continue to be available on the IT Jungle site. This list of books is proof of the fact that the WDSc has been a work in progress and is still not complete. The GUI Designer that is missing from the Standard Edition right now is also missing from the Advanced Edition. It has not been built yet. So, while your work is in progress IBM, do things to intrigue the masses to join the party and use your wares. Charging a tax, especially at this stage of the product life, appears to your public very clearly as a usury proposition.
So, the bottom line is that we have a software set that now, though not perfect, has enough good stuff and enough close-in plans that IBM can actually brag about it. But, it took a long time to get here. Now that it has, the Software Division has rained on the System i parade. History repeats itself. The last time Dave Slater, the former head of marketing for IBM’s Toronto labs, said that everything is there and it is free, Software Group cancelled the free WebSphere Server and told us that we needed to spend $12,000 on something called “WebSphere Server Advanced Edition.” Does that sound familiar?
It took IBM three years to recover and it killed all the work that Slater had done to stage the System i client set (that means all of us) for a triumphant entry into the world of the Web. I think the Software Group just did the same thing again on February 27. Either we ought to secede or we ought to get rid of those guys. We can’t afford friends like that.
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 Consulting. He is the author of 27 books, has written numerous articles about current IT topics, and is a frequent speaker at the COMMON and other technical conferences and user group meetings across the United States.