The Pizza Disconnection
Published: June 5, 2007
by Hesh Wiener
On May 23, when most people were thinking about the next weekend's holiday barbeque, in Minneapolis, Federal Judge Richard Kyle was preoccupied with pizza or, more accurately, the tallying of various slices of the pizza trade. Before him was a case involving Domino's Pizza and a number of its franchisees. Domino's insisted its franchisees buy some IBM hardware it resold plus a unique software suite. The franchisees said No Dice, or maybe No Tiles. The franchisees won. Their deal with Domino's, the court said, allowed them to shop around for hardware and maybe for software, too.
The case wasn't immediately significant for Domino's, whose stock ticked up as the story broke the following day, according to the Dow Jones wire service, but IBM shares lost a buck and change, very possibly because of the court case. If the Domino's deal had worked out and the court case gone the other way, it would have given IBM a way to turn franchisors into resellers that could boost its margins in server sales. But the way things turned out, the Domino's franchisees will be able to buy alternative hardware or to buy IBM hardware from a dealer who wants less for the stuff than Domino's was going to charge. The implications of the case are clear if the underlying hardware was X64 iron, which can be acquired from many sources, but the issue is not limited to commodity servers. The case could also pack a wallop in the mainframe business, where IBM exercises considerable control over hardware prices, at least for the moment.
The Dow Jones account of the case makes it clear that the judgment was based on contract law, not antitrust law. The dispute was about what Domino's could compel its franchisees to do and what it could not. Nonetheless, the ruling appears to break what might have been an attempt on the part of Domino's to illegally tie hardware sales to the sale of a software package. The court held that the most Domino's could demand from its franchisees is that they comply with any technical requirements it sets down to standardize the reporting of financial information by the franchisees to the franchisor. Obviously, this requirement can be met by software and specified communications technology, no matter where the hardware comes from. The court said that Domino's demands were not about the pizza business, only about the dough. So, in the end, some franchisees will buy the whole pie from Domino's, while others will only take a slice of it.
In the mainframe business, the franchise of interest is not about selling pizza but about selling System z boxes and their software. IBM's authorized dealers, who officially go by the name business partners although it's a stretch to think of their relationship with IBM as a partnership, operating under a lot of rules. Among other things these rules make it difficult if not impossible for an end user to get bids on a mainframe from more than one dealer. While we cannot provide all the details of IBM's arrangements with dealers, which are jealously protected by IBM and fearfully protected by the cowed dealers, the general way things work is not a secret at all.
In order to get a System z from IBM at a nice discount, which translates into a nice profit for the dealer, the dealer has to have a unique relationship with the customer. The dealer has to spend a certain amount of time face to face with customer personnel, to do a serious job of preparing a proposal that matches the customer's needs and to perform a number of other tasks that add to the quality and quantity of the dealer's sales effort. This is good for the customer, good for the reputation of IBM, and good for the reputation of IBM's dealers.
But the way the rules are written only one dealer can get a top-notch hookup with any particular customer. A second or third dealer who wants to bid has to pretty much get the first one tossed out and then build a new exclusive relationship with the customer. This is not a common occurrence. Without this special relationship, any dealer that wants to pick off a customer must take a significant financial hit in the form of a drastically reduced dealer discount from IBM. In other words, a second dealer might have to make little or no money on the system, or even take a loss on the hardware, to match the price at which the established dealer at a customer site can sell the iron.
The impact of this on the dealer ecosystem is good and bad. A smaller dealer is to a considerable extent protected from predation by a larger dealer who might be willing to work on a smaller markup in order to grab the customer. On the other hand, without any real competition the customer never knows if it is getting the best possible price on hardware, or even close. Every deal is done in secrecy. A lot of deals involve complicated discounts based on the customer's requirements and growth plans. Some are not on the list of scenarios IBM provides for its dealers, a list that covers many but not all sales situations, and these end up becoming what are called special bids, deals in which IBM and the dealer work out a price based on the specific sales case. If the results if an extra 5, 10, or 15 percent price break for the dealer, the customer is unlikely to ever find out, and will remain in the dark about how much of a price break the dealer got and how much of that was passed through in the dealer's bid.
Now, IBM is doing a pretty good job selling mainframes these days, and so are its dealers. That means the market is bearing the prices asked for z9 boxes and their associated software. But it doesn't mean the customers are happy paying as much as they do. It only means that prices are not so far out of line that they are leading to a mass defection . . . or that if there is a significant loss of customers, it's more than matched by increased sales to the customers that are sticking with the IBM mainframe.
As it turns out, there is a portion of the mainframe base that's a bit less happy with the situation than the overall figures from IBM might suggest. These more tentative customers are generally smaller shops, shops running VM and VSE operating systems, and shops that are perfectly happy with their old installed systems who feel they are getting shoved toward new machines and new software that they really don't want to buy and wouldn't buy if IBM didn't make it clear that support for its old technology was going to be phased out.
This small, dark corner of the mainframe market happens to include a lot of small software developers who, until recently, ran IBM software on Flex emulators and not real IBM hardware anyway, and IBM and its dealers are not inclined to be worried about breaking their hearts or, for that matter, their budgets. IBM offers developers remote processing under terms IBM thinks are pretty good, an opinion that might or might not be shared by all the developers. The commercial users in this corner of the mainframe world probably don't need whole mainframe either, and IBM and quite a few other companies offer hosting services that can deliver mainframe MIPS by wire for a lot less than the cost of a whole standalone system.
Nobody out there is completely bereft of choice. It's just that some of them understandably feel there ought to be just one other kind of choice, the kind that comes about when a number of dealers can compete for the user's business on a level playing field. And, the way IBM's dealership rules work out, there is no level playing field when it comes to hardware sales. (The software deals users get are much more uniform, whomever the dealer might be, even if they are governed by complicated pricing schemes.)
Users who decide to unplug their mainframes and instead look for mainframe hosting companies find that there is a lot more real competition than the ones who feel they absolutely have to have their own hardware. If these users feel, for security reasons or for other reasons or for no reasons at all that they should own and operate their own mainframes, they end up with Hobson's choice.
If the rebellious spirit of the Domino's Pizza franchisees infected the mainframe base, things would change, and, as far as some users are concerned, change for the better. With a little more competition one user might end up getting a new mainframe but a storage expansion based on used disk arrays, while another might get a memory bump at one-third off list. Sure it takes a spreadsheet the size of Kansas to turn the specifics of a mainframe deal into even an approximation of the total cost of ownership of a system for, say, a three-year period. But this, it turns out, is not much of a burden for customers. Any mainframe dealer will happily build a spreadsheet model that covers pretty much everything in the budget. A customer who gets two or three different models from different dealers can then compare them, figure out which models seem most plausible, and end up with hard numbers to put into the purchase decision along with the soft factors like the character, proximity, and technical viability of various prospective suppliers.
Dealers might hate this. They might feel the way a car dealer does when somebody comes in for service on a car bought elsewhere on the cheap with the help of the Internet. But it would be foolish to underestimate the cleverness and tenacity of IBM's mainframe business partners. They already live under a set of rules that could give a contract lawyer a heart attack, if you happen to know a contract lawyer with a heart, and, for the most part these dealers are making a comfortable living. We also suggest that any changes in the way the mainframe trade works that give customers a little more choice would not undermine the quality of the equipment, software, and support customers receive. IBM is simply would not let things slip, and if more of its dealers worked up a sweat a bit more often, that would just become part of the way things work.
Everyone on the selling side apparently thinks mainframe customers are too complacent to act the way the franchisees in Minnesota did. We're not sure that's really the case. More likely, they just don't quite know whom to call. Maybe they should try Domino's.
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot