For Some, Entry z9 BC Models Are Cheap

Published: July 11, 2006

by Hesh Wiener

For the astute shopper, or the lucky one, the entry systems in the z9 BC line can be very affordable. Under the right circumstances, and by ordering the right configuration, a customer might get a quarter to a third off suggested list price. But not everyone is going to get a great deal. To get a killer discount, a customer has to be in the center of IBM's target, and the seller (a dealer or an IBM suit) has to get IBM to approve the price break.

Currently, the best z9 offers are aimed at mainframe shops that are still using Multiprise systems, either Multiprise 2000 or Multiprise 3000 models, who move to one of the three smallest z9 BC models. Multiprise machines are just about the only IBM mainframes still in the field in significant number with very low processing power by current mainframe standards. To get a deep discount, the buyer would have to be interested in a 26 MIPS z9 BC model A01, a 38 MIPS model B01, or a 46 MIPS models C01. Multiprise shops that want to move to a z9 with more than 46 MIPS can still get nice discounts, but they are not the main target of IBM's most aggressive pricing plan.

Special price breaks are not entirely confined to customers moving ahead from a Multiprise. IBM will cut hardware prices for users who want to bring in an additional mainframe in the form of a low-end z9 BC. These add-a-server deals are similar to the ones aimed at Multiprise shops. But the deepest discounts are not automatic. Buyers have to fight for them, and the resellers that buyers turn to may have to carry the battle to Big Blue. IBM must approve special deals made by its resellers, more or less the same way it approves special deals offered directly by its own sales force, and sometimes the company resists. IBM does not want to undermine the standard pricing scheme its marketing people have developed.

The details are hard to pin down. People who are in the know are understandably reluctant to talk about topics they consider inside baseball. What seems to be the case is that resellers (and IBM reps) usually don't have to jump through hoops to close a deal because IBM has done most of the work already. The z9 BC at list price (or close to list price) is inherently pretty attractive to most prospects. That means the buyer has to make a credible case that the mainframe purchase will hinge on a price cut.

The shops that are likely to have the hardest time getting a price break are those that have already moved to 64-bit technology in the form of a z800 or z890. There are very few z890 users who would have a reason to change to a z9; and the ones that do are probably thinking along the lines of z800 users who find the z9 attractive. These shops want to move to the z9 because it has more granularity and in particular they want to move to a machine with a lower MSU rating to cut their software costs. IBM doesn't welcome downsizing, but if a customer is serious about ditching the mainframe for a different platform, even the toughest rep or reseller may offer pricing incentives. Even so, because users with z800 or z890 systems have recently made big investments in mainframe technology, IBM may correctly regard any attempts by these shops to threaten defection as largely theatrics. But there are clear exceptions. If a mainframe shop is part of a company acquired by, say, a firm with central systems running Windows, Linux, or Unix, IBM will see that its presence is really under threat and it will react.

Whatever the customer's situation, the key to getting a lower price is knowing what is possible. And what is possible depends on what IBM thinks is a vital target and how IBM believes it can hit that target. In the case of the z9 BC, IBM's emphasis on the entry models indicates that Big Blue doesn't want to see the bottom of its mainframe base erode as it has in the past. Between 1998 and 2005, according to IDC, IBM saw its mainframe installed base fall by more than a third. A lot of the lost footprints can never be regained. They were not the result of consolidation but rather the consequence of migration from mainframes to other platforms, including IBM's own AS/400 and iSeries as well as Unix, Linux, and Windows servers from IBM and others. The largest mainframe shops, for the most part, stuck with Big Blue, but users whose mainframe applications required only modest processing power found the temptations of migration, real or imagined, irresistible.

IBM belatedly realized that not only were these small shops were worth keeping, but also that the small shops that experienced successful migration were potential models for larger shops. The result was that IBM started to offer machines that better matched the cost and performance requirements at the low end. The two key IBM tactics were the introduction of low-powered mainframes, such as the derated z890 models, and the development of usage-based pricing schemes that made software for small platforms more affordable. (The next big move will most likely be a transition to Power processor technology that will consolidate to a significant extent the engineering and manufacturing processes used in mainframes with their equivalents in the System i and System p servers, but that's not quite on the horizon yet.)

In order to defend its base at the vulnerable low end, IBM has become unusually eager to please shops that use only a portion of the power of the smallest Multiprise 3000, the 60 MIPS H30, or one of the single-engine Multiprise 2000 machines, of where there are quite a few models. These shops are already on a knife edge, and they could tip toward a mainframe future or away from it. To retain these users on the edge of the base, IBM is now willing to slash hardware prices . . . but only in cases where the customer is serious about making choices that include alternative platforms. Additionally, customers have to know a little about the nature of the deals IBM is willing to make in order to obtain the best possible bang for that budget buck.

First, the customer has to be near the center of IBM's target, which at the moment means a user who has an installed Multiprise. Then, the customer has to want an entry model z9 BC, specifically a model A01, B01, or C01. Finally, the customer has to be willing to take a machine that's a bit larger than the base model. This last point isn't obvious, and it might mean that some customers really can't get into the optimal position for deep discounts.

IBM is willing to cut hardware prices so a user can save about a third of the cost of a model A01 and a quarter of the cost of a model C01. B01 deals will fall between these end points, most likely closer to the C01 end of things. But IBM also does not want to sell an absolute base system. IBM wants to sell machines that have extra features. Users must understand that these features are expensive, but in the end they might in fact not add to the cost if the alternative is to acquire a skinner machine without much of a price break. Shops that can use extra memory to boost throughput might get into the center of IBM's target by adding an addition 8 GB to their base system. Others might want to add channels or OSA Express network cards. None of these features will affect maintenance prices, so the cost is a one-time expense. Users who can get more work out of their systems by using the extras will end up getting a bit extra for their hardware and software dollar, making the larger configuration somewhat easier to justify to management.

I have tried to put some hard numbers on the situation, and it's all in a nice little table here. To get the deepest discount, a customer has to add features to the base system so that the total cost before discounts is above the "real" minimum, the lowest price IBM wants to actually get for a mainframe, whether that price involves a discount or not. (It may be possible for a customer to get a special deal at a price below what I call the "real" minimum, but it is my understanding that IBM has told marketing folk that it very much wants them to respect its minimum target price guidelines.)

Here is a simple example. IBM wants to sell A01 systems for at least $135,000. A shop that wants an A01 might think that it can get a box for about the $100,000 that IBM said was a starting point when it announced the z9 BC. It can, but it won't get much of a discount. But if that same shop adds some FICON connectivity, additional network adapters, and so on to the base box and brings the price up past $135,000 then it's possible to knock the cost of the deal back down to about $100,000. If the configuration is larger, the chance is even greater that the customer will be able to wrestle a third off the suggested list price.