Computer hardware and software

MARKET OVERVIEW

The summer period was very slow for all segments of the IT market, which is quite normal as most of Europe disappears on holiday for July, August and much of September, and even the US takes longer holidays these days. In addition, with the increasing 'globalization' of large organizations, decisions that affect only the US are today often subject to non US influences, which makes the long summer holiday period far more of a worldwide issue.

However, as winter arrives, activity is beginning to increase in all areas of IT and in particular the mainframe market. This was particularly noticeable as IBM pushed very hard for hardware business toward the end of September and in at least some cases this resulted in equipment being 'shipped' before the orders (or certainly the finance agreements) were actually signed--reminiscent of the old days before accounting scandals such as Enron.

As this was mainly a hardware push, some of the irrational behavior of the users was apparent to an even higher level than in previous periods. This is because although things move slowly in the mainframe area as will be explained later--the actual economics change far more rapidly. For example, in 1980 the processor hardware purchase price of a system was far greater than the five year IBM software cost--indeed in most cases it was greater than the IBM and ISV software cost together. By 1990 however, the hardware purchase price of a system was now similar to the five year IBM software cost and today, for most users, the IBM software cost over five years is 15 to 20 times the hardware purchase cost. The combined IBM and ISV software cost in many cases is in fact a staggering 30-40 times higher than the hardware cost over five years.

The change in the relationship between hardware and software costs in the storage area is not as dramatic but, as detailed last month, it is moving in the same direction.

Yet users seem to have almost ignored this change in their decision making and continue to sign quite outrageous software contracts (IBM and ISV) almost without any negotiation or approval process--the defence being that there is little they can do about these costs and they are ongoing anyway. In fact it is the latter point that suggests they need more careful evaluation and negotiation. Yet still they deliberate for weeks if not months and then seek board approvals for what are by comparison relatively insignificant hardware purchases. This is despite the fact that such hardware purchases are often 'free' when considering the software and maintenance savings.

One recent example was where a user signed a five year IBM software deal for over $50m after just a few days deliberation, yet decided against a processor order for $3m without full board approval. This was despite the five year software and maintenance savings of this particular hardware acquisition fat exceeding the $3m hardware investment required.

Of course the software deal also showed major savings, but the level of commitment was still almost 17 times higher.

This problem, repeated in many users recently, is largely the fault of IBM which largely does not produce detailed business cases to support hardware purchases today. This fact has been verified by reviewing numerous IBM and IBM business partner proposals and suggests that the vendors need to get back to basics and illustrate how a new hardware acquisition is not only cost effective, but will reduce the ongoing costs of the user. For example, in most cases the IBM processor sales team seem unaware of the maintenance savings from a move to the z990 from the z900 systems and do nor have a figure to hand for the software savings offered by such a move.

Yet the business case for replacing a z900 with a z990 can be very simple as in an example negotiated recently where the final negotiated costs resulted in a 13% lower total annual cost and a 42% lower annual cost per Mips.

As the new systems required no more people to support them, actually somewhat fewer, and less physical space and power, the overall savings were even greater than detailed above.

Such an acquisition fully supports a user's need to both lower his total costs, a not unfamiliar situation today, as well as providing a platform to introduce new applications or port applications from other platforms to further reduce overall costs.

Apart from 'bad' software contracts (both IBM and ISV) that can influence these figures, but which can be re-negotiated relatively easily, the only other problem with delivering this level of benefit to all users is the IBM z990 manufacturing methodology, using four 'books' each of eight processors. This continues to cause problems for IBM in the field because where a user is expected to grow quire rapidly into a larger model than initially required, for example from an A08 to a B16, IBM tends to bid the larger model from the beginning even though this typically adds $300,000 to $500,000 to the purchase price. It also adds significantly to the maintenance cost as will be seen latex all to no obvious customer benefit.

Of course sometimes the need for the larger 'book' to begin with is essential but only created by the requirement for 'special' processors such as a CBU, IFL, zAPP or ICF--but even here this is a problem of IBM's own making. By allowing a single processor engine to support multiple such processors in some partitioning mode the problem could effectively be removed in most cases.

This 'engine' problem was highlighted in a recent case where a user wished to upgrade to model 306 z990s for economic reasons as explained in earlier articles, but actually only use 30 models initially. He needed three CBUs as well in each system pushing him over the limit of eight processors in the A08. IBM it seems was unwilling to let one of the two 'idle' engines from the purchased 306s be used as CBUs until the full 306 power was needed and a B16 upgrade actually required. Instead IBM insisted on the systems being B16s from day one. The result was that IBM, at least temporarily, has lost the order.

SOFTWARE

When discussing IBM mainframe software many users think either only of MLC or possibly of MLC and OTC products. However it is today far more complex than this as detailed in the illustration 'IBM pricing model' opposite. Sitting on top of everything is normally an ESSO or ELA agreement for three to five years. The construction of these has been detailed in previous editions, but the significance is changing. A few years ago to the mainframe user the MLC element represented some 80-90% of the mainframe costs for most users, with IPLA of little significance at no more than 10-20% of the mainframe cost. Passport Advantage was typically being handled by another department and of little concern to the mainframe organisation, although often accounting for at least as much expenditure as the MLC software.

Today, however, the structure of an EIA/ESSO dictates that the user understands the Passport Advantage area as the OTC commitment here can contribute towards lower MLC costs and, in future, the same may be true even for the S&S element of both the Passport Advantage and IPLA products. For users currently negotiating long term IBM software contracts such as an ESSO, ELA or OIO, it is important to understand that as these relationships and amounts change in significance, IBM must introduce some new rules soon to cover the users' needs for the next few years. For example, both the Sysplex issues where the size of the current systems has made today's rules inoperable for many large users and other issues such as the growing OTc inventories and related S&S payment streams require that IBM makes changes.

The second of these is, in particular, already causing major issues for many users who signed ESSO/ELA agreements, but are finding that the MLS Caps are higher than needed but the OTC commitment (IPLA and Passport Advantage) too low--with an obvious knock-on effect for the S&S. They are thus faced with paying for MLC capacity within their deal that they don't need, while having to pay additional amounts for 'new' OTC products and S&S, which are frequently just new versions of older MLC products containing new (but essential) functionality.

Exactly bow IBM will handle these changes remains to be seen, but you can be sure that your local IBM team will take months to understand and exploit the changes to your benefit. During the changeover this will create many problems for the user.

MAINTENANCE COSTS ON THE Z990

One of the major issues on the z990 is the cost of the 'books' that as noted above can distort the purchase price significantly. But it does not end there, the 'book' approach also impacts the maintenance price significantly.