Analysis: New AS/400s Supports Better Development

Mark Buchner

In the coming months, IBM's focus will be on emphasizing the relative advantages to be gained by customers who choose IBM hardware. The promotion associates all IBM server platforms (AS/400, RS6000, Netfinity, S390) with the tag line "magic box."

From a developer's perspective, the AS/400 magic box, (today's 170, 720, 730 and 740 models), is significantly better for the support of AS/400 developer's needs than the boxes they replace. Let's take a look at why there have never been better AS/400 machines to support development activities.

Start by going back in time.

Years ago, the AS/400 basically supported only one style of application development, the catchword for which was "native." This meant it used a highly proprietary system covering RPG/CL/DDS. From a hardware perspective, the machine was architected to support that development environment.

The old CISC AS/400 could deliver excellent, diverse I/O related performance, but it was the result of using diverse (slow) complex processors, not a single fast one. Newer programming models (such as C, Visual programming, Object Oriented languages and Java) require fast processors. Today, so-called "threading" techniques (separation of work units among processors), are coded into applications as an afterthought. A key reason why new application development paradigms did not quickly catch on in the AS/400 community is because the CPU was too slow.

More recently, along came RISC, which fundamentally reengineered the CPUs of the AS/400 to be 64-bit and PowerPC-based. After a few generations, the advancement of this new technology resulted in the AS/400e series (Model 6xx, Sxx, 150 and 170).

These models finally provided the raw price/performance and scalability to allow the most advanced development to perform competitively with any other platform. The models provided backward compatibility, or "investment protection" for existing software.

Therein lies the dilemma with these models. IBM provided one set of AS/400 hardware, the "servers" (S10, S20, S30, S40, 150, 170) which delivered competitive modern application development boxes. Another set, the "systems" (600, 620, 640, 650) provided investment protection.

For the servers (red boxes) to be competitive with other hardware platforms, the boxes sold from between $130 to $200 per CPW. The compromise of these models is that while they aggressively support new, open development paradigms, they punish their users for simultaneously (or alternately) using the old ones. This compromise was achieved through the use of confusing "interactive" and "batch" CPW ratings and performance algorithms.

The system models (blue boxes) supported upgrades and software investment protections. Customers could freely mix any kind of application, old, new or modernized. But the compromise is the cost--between $500 and $1,000 per CPW--on average, four to five times more than the red boxes. For comparison, a model 620 2180 delivered 113.8 CPW for $115,000. An S20 2161 also delivers 113.8 CPW, but for $24,000. Its interactive capability, however, is restricted to 31 CPW.

Customers eventually demanded to know more about these price differences, but especially why:

1. The existing AS/400 customer who upgraded to "e" series was almost discouraged from modernizing their application because of the dollar per CPW premium placed on system boxes. At best, they were encouraged to distribute their processing by purchasing a new machine for new applications and development.

2. The new AS/400 using modern applications and development methods was restricted from integrating to the installed application base because server boxes discouraged upgrades. Given the AS/400 attraction as a consolidation platform, this pricing seems to be a step in the wrong direction.

To my way of thinking, the existing AS/400 customer wants three things:

1. A platform to provide investment protection for existing, proven applications.

2. A platform to modernize, extend and coexist legacy software with new technologies.

3. The ability to do both 1 and 2 on a single, consolidated platform.

This year's announcements, culminating in the August announcements of PCI Towers and fast disks, mean customers now have precisely what they wanted.

Now, the customer chooses the CPW necessary to run existing applications and receives a huge dollop of client/server CPW to support the modernization of applications. There is still a premium for using interactive CPW, but this is paid for directly only when needed (through interactive feature cards) and without sacrificing consolidation opportunities.

The AS/400 customer will continue to a pay a premium for the protection of old software investments. But they will not pay any premium when it comes to using the AS/400 for new application development.

The AS/400 "magic box" brings the best of the Old Development World and New World onto a single consolidated base.

One caveat: The yellow AS/400 "killer bee" boxes are a new variant. These are for customers who want to use the AS/400 for nothing more than serving Domino applications. In order to get this particular feature, the customer forfeits the integration and investment protection benefits of the AS/400.There is an additional dollars per CPW reduction from the server models and customers need to be aware of this before they make the investment.