Mad Dog 21/21: iBird In The Hand Or iToo In The Bush
April 3, 2017 Hesh Wiener
IBM i shops seem to be pretty content with their systems. Profiles of the installed base, such as that recently published by HelpSystems, show a mix of old and new equipment with an impressive span of performance. For users, the stability of the base is sign that owners of the systems are satisfied. But for IBM, ISVs and other suppliers, an i that is not evolving is a dead duck. Suppliers to the IBM i market want the platform to become a phoenix, a fresh system rising from the ashes of aging predecessors.
In the past month, we have broke the news that IBM was introducing a Power Mini platform that could give the i a fresh start. It is sufficiently small and affordable to serve as a gateway to the IBM i realm. It is nevertheless capable of supporting workloads greater than those possible on the oldest and least versatile installed machines. Consequently, it provides an attractive migration path for users of low end installed systems as well as a starting point for a new or supplemental installation.
It remains to be seen how easily legacy applications can be moved to the new system, which must run IBM i 7.2 TR6 or 7.3 TR2 or later. Whether that systems upgrade will be pleasing to users of older systems will depend not only on IBM’s migration support for user-developed applications but also on the level of interest and commitment evidenced by key ISVs.
The history of the IBM i, from its origination as the System/38, through its various AS/400 incarnations and forward into the i era includes several major transitions. IBM has generally done quite a bit to help users move ahead, but some of the leaps, when processor hardware was substantially changed, presented a daunting challenge to end users. This was particularly the case at small shops that had application programmers and operators but few in-house systems specialists if they had any at all. Eventually, the most challenged users had to choose between moving ahead and moving away. That second option has generally been a one-way street. Customers who leave IBM’s proprietary business system almost always migrate to an X86 platform, where fierce competition in hardware and the lively rivalry of Windows Server and Linux keep vendor pricing under pressure. On the other hand, IBM i applications and middleware, not to mention the operating system itself, are not supported on X86 hardware. So, while there may be financial benefits available to some shops that can move from an IBM i to an X86 alternative, there are also costs, including education costs. The barriers often turn out to be higher than customers expect.
One of the alternatives that increasingly may tempt users who feel they can move of their IBM i work to a different platform is the emergence of service companies, cloud computing firms if you prefer that term, with offerings aimed at IBM i shops keen to preserve portions of their workloads even as they switch the bulk of their production to X86 platforms. Sure, users can retain some i equipment even as they move the majority of their work elsewhere, but for small companies the preservation of staff to support disparate environments is not affordable. Users who want to unplug their IBM i machines but not all their i workloads often turn to service providers. IBM could, in theory, become the go to cloud i provider, but IBM has not yet developed service offerings for very small i workloads, so users are obliged to turn to third party providers.
This situation is similar to that in the mainframe world, where small mainframe shops fell out of IBM’s target area as Big Blue’s big iron grew ever more powerful and expensive (even if price/performance has continually improved). Companies that have small but vital mainframe workloads or a need to retain legacy data and applications for tax or regulatory compliance may unplug all their mainframes and move their indispensable residual work to third party firms that provide remote hosting. IBM is a prime candidate for large remote computing service arrangements, but it cannot find a way to gear down its mainframe offerings to customers with very modest needs.
Even these days, with IBM offering cloud services through SoftLayer and what it has grown into, similarly lean and affordable offerings tied to IBM i and mainframe platforms are not viewed by IBM as a strategically important service. There doesn’t seem to be a strong business case for this kind of operation within IBM’s cloud services framework. IBM sometimes talks about diversifying its remote service offerings to better support the wide range of computing shops that use IBM platforms, but users who look into the matter seem to end up with third party services.
The provision of a new, economical low end IBM i system is thus a game changer. (So, too, would be a tiny IBM mainframe, but IBM tried that a few times and in every instance it dropped out when the numbers didn’t impress IBM management.) The Power Mini gives IBM i shops itching to move off the i a new reason to stick with the system that has served them until now. If the new machine provides good support for legacy applications and attractive migration offerings from IBM and ISVs, companies that were moving to X86 for most of their work might be able to sustain vital remnants of their i operations without going to a third party service company. Moreover, some third party firms might be able to use the new, lean IBM i as the basis of an offering for users who didn’t want to staff up for i support but want a whole, small i systems managed by a specialist hosting company.
While we have gathered up pricing on the Power Mini, we still don’t know how Big Blue and its partners will price related products and migration services. If IBM was really keen on using the new entry i as a way to herd the most severely lagging companies in its base toward contemporary equipment, it could probably come up with migration offerings that users would be hard pressed to decline.
To win over the laggards in its IBM i user base, IBM would have to make the cost of acquisition and ownership attractive for at least five years, perhaps longer. If IBM’s offering includes a no-cost or low-cost technology refresh along the way, the full package could be built into the initial deal price. Prospects would see that IBM wants to provide a nice future for users of smaller and older systems, something that it hasn’t done in a while. In the past, IBM had entry systems such as the Model 515 of a decade ago that were sold in bundles with software and support. When IBM offered that kind of product, a lot of dormant users woke up and refreshed their systems.
The cost to IBM of providing the low-end i server is peanuts. The machine is a variation of the Power S812L Linux box, pretty much eliminating hardware design costs. The software is a tailored configuration of established production products that provide a single image virtualized operating environment. Nothing about the Power Mini suggests IBM will have to make much of a fresh effort to deliver middleware support.
Now that the foundation firmware and software is in place, higher layers will be similar if not identical to software used on contemporary machines. In that way, the IBM Power Mini running i resembles standardized X86 hardware running Windows Server or Linux. On X86, once the BIOS, operating system, and key hardware drivers are in place, most of the code layered on top will run on all compliant boxes of any particularly generation. And the farther from the iron an element of software is placed, the wider its applicability.
What IBM might want to do is not tailor its hardware and low level code to make migration easy, it might provide hands-on support for actual migration. IBM would have to come up with tools and even personnel who understood how to move workloads from a particular type of legacy i to the new machine. And then IBM might have to give this technology and service offering away at no cost, to bake it into the acquisition deal in a manner that makes customers understand that the whole move, from an ancient i to a current one, can be a bargain if the user organization follows a path mapped out by IBM.
There’s no question that users would make the move as long as the costs were low and they did not have to sacrifice familiarity with their old applications and production procedures. Whatever additional features the new machines provide, they would also have to do a great job of running the work that slow-moving (or immobile) customers have on their old boxes.
Sure, there will be cases in which migration is difficult. But if IBM really does its homework, the number and severity of nightmare upgrades would be small – miniscule in the context of the multi-billion business surrounding IBM’s i and other Power systems. And these days, with IBM long gone from the X86 hardware world, the boost to Big Blue would be very important. IBM has nothing else to offer i and Power shops who want only a small helping of contemporary computing power. It must offer a compellingly superior system, first rate in technology and in value, or face the loss of its smallest but most loyal customers.