January 20, 2009 z/OS competition could save Europe $48 billion over 20 years
By Jeff Gould, CEO & Director of Research,
Peerstone Research
Outline of Peerstone’s approach to the European mainframe market
Peerstone Research examined the economic costs to Europe of IBM’s mainframe monopoly. The point of departure for our work was the fact that thousands of Europe’s largest companies currently run their most critical business computing applications on IBM mainframes at very high costs, without the benefit of a competitive market for mainframe hardware and software. Our research aimed to answer the question of how European companies and the broader European economy would benefit if a competitive market for computer platforms capable of running these highly valuable “legacy” mainframe applications were allowed to develop.
We looked at the issues from several angles:
- How much would European users save if IBM agreed to license its proprietary mainframe software (currently restricted to IBM’s equipment only) for use on alternative computer hardware?
- How much would European users save if, in addition to competitive pricing for mainframe hardware, they were able to benefit from competitive pricing for mainframe software and related professional services?
- What options do IBM mainframe customers have for switching to other platforms, and how feasible are these options?
- How do IBM’s proprietary mainframes compare to servers based on industry standard Intel or AMD microprocessors?
Mainframe hardware competition could save Europe $48 billion [1] over 20 years
We compared the costs of current IBM mainframe hardware systems to the costs of high-end servers based on modern Intel or AMD microprocessors capable of running high-value business applications at equivalent performance levels. We found that:
- The introduction of a fully competitive market for mainframe hardware would save European mainframe customers $48 billion over the anticipated 20 year remaining lifespan of current mainframe applications.
- These savings would result from the decline of current monopoly mainframe hardware prices to the level of prices for standard non-mainframe servers of equivalent performance (e.g. high-end servers from leading vendors such as HP, Sun or IBM itself using modern Intel or AMD microprocessors and running modern operating systems such as Linux, Unix or Windows Server).
- We found that such industry standard servers would be capable of running existing high-value “legacy” mainframe applications with the same performance as IBM mainframes but at a small fraction of the cost (typically less than 10% and in some cases less than 5%).
- We further found that IBM sells its own mainframe hardware at similar dramatically lower prices to a restricted subset of customers who wish to run only new Linux or Java applications, but deliberately disables these discounted mainframe processors in order to prevent them from running “legacy” mainframe applications.
Mainframe software competition could bring total European savings to $100 billion
Since prices for essential mainframe software such as operating systems and middleware are typically far higher than prices for equivalent software running on industry standard servers, we sought to determine how much Europeans would save if prices for mainframe software were allowed to fall to competitive market levels. We found that:
- A fully competitive market for mainframe software would result in additional savings to European mainframe customers of roughly $50 billion over 20 years.
- The total savings to Europe from a fully competitive market for mainframe hardware and software would be on the order of $100 billion over 20 years.
Migration is not a realistic option for European mainframe users who wish to move their high-value legacy applications to cheaper platforms, but emulation – although currently forbidden by IBM – is highly effective.
We looked at the options mainframe users have for moving their high-value “legacy” mainframe applications to cheaper computer platforms, and found that there are two broad approaches to this problem:
- Migration, which means translating the code of an application originally designed to run on an IBM mainframe under an IBM operating system into code capable of running on an industry standard server under a modern operating system such as Linux, Unix or Windows Server.
- Emulation (or, more technically, mainframe instruction set virtualization), which means using specialized software or firmware to “emulate” the IBM mainframe hardware on an industry standard server, and thereby allow high-value legacy mainframe applications to run on such servers without modification.
We compared these two approaches and found that there are major differences in cost and technical risk between them. Specifically:
- Migration of legacy mainframe applications to other hardware platforms and operating systems is forbiddingly expensive and risky in most cases. This difficulty is caused by fundamental technical differences between mainframe system software (such as z/OS and CICS) and modern system software. Successful migrations of legacy mainframe applications are rare and probably account for less than 1% of the mainframe installed base in any given year. Most so-called “migrations” are in fact expensive reconstructions of the original legacy application on a new platform.
- Emulation of the mainframe instruction set on non-mainframe servers based on Intel or AMD processors, by contrast, is a proven and effective strategy for moving legacy mainframe applications and their associated IBM system software to these cheaper hardware platforms. Many methods with varying performance and cost characteristics exist for performing emulation (which is a form of virtualization). However, IBM’s restrictive licensing practices currently make it difficult for customers to benefit from these innovations, because IBM refuses to license the operating system software such as z/OS required to run legacy mainframe applications on emulation technology.
Although IBM uses restrictive licensing practices to prevent customers from moving their mainframe applications to less expensive platforms, IBM mainframes no longer offer intrinsic technical advantages over modern server platforms.
We compared the architectural features and performances of late-model IBM mainframes with current high-end enterprise-class servers built with industry standard microprocessors based on Intel’s x64 architecture. We found the following:
- Modern high-end multiprocessor enterprise servers from vendors such as HP, Dell and IBM, based on multicore x64 microprocessors such as Intel Xeon or AMD Opteron, have equivalent computing power to IBM mainframes when augmented with appropriate emulation technology.
- Although IBM’s restrictive licensing policies prevent customers from shifting legacy manframe applications to these modern server platforms, the mainframe possesses no inherent technical superiority over appropriately configured industry standard servers of equivalent performance.
Author’s Biography
Jeff Gould is the founder and CEO of Peerstone Research, a San Francisco-based IT industry consulting and market research firm. He is the author of numerous publications about enterprise software and hardware. Prior to founding Peerstone in 2001, he served as Editorial Director for the European operations of American IT publisher CMP Media, where he oversaw the European editions of InformationWeek and Computer Reseller News. Prior to joining CMP Europe in 1994, he was the founder and CEO of Datastrategies SA, a Paris-based IT market research firm specializing in surveys of European corporate IT users. He has extensive experience in all aspects of enterprise computing in both Europe and the United States, and is a UK-US dual national. He can be contacted at jeffgould@peerstone.com.
Download the research report PDF (248K)
[1] We use dollar values instead of euros throughout this paper because the IDC server market sizing data we refer to is expressed in dollars.
Reader Comments (4)
I think the idea of open computing is a great idea... portability of application code to different platforms is obviously a good thing. However, to characterize z/OS just running on a IBM hardware only as a monopoly strikes me as a little silly. There plenty of options available for a companies processing needs. Competition should be judged as available alternatives which clearly exist. If a company feels the total cost of the mainframe is not delivering the lowest transaction costs, there are many methods to migrate and lots of companies that help with this transition. Markets cannot be defined so narrowly. The competition for television viewership is not just between satellite and cable, rather its all alternatives for entertainment. Demands on a company to relinquish its competitive advantage seem to me to be interventionist. IBM has shown a willingness to help clients get more value from open computing with their investment in Linux, IFL's, and eclipse. Lastly, isn't SOA/web services/cloud computing, etc. changing the economics anyway? I think it is a bit laughable to estimate any savings as large as you suggest that far in the future.
Jeff,
Given your background, "extensive experience in all aspects of enterprise computing", and being the CEO of a research firm, one would assume you would have done your homework here. Clearly you have not.
Your argument strictly challenges the cost of acquisition. Which, from that narrow viewpoint, your argument is sound. However there is a broad spectrum that needs to be considered when you coin the phrase "Enterprise Computing".
I would argue that in your 20 year view, European IT costs would skyrocket to 3 - 4 times today's present levels embracing an x86 deployent model. Possibly higher. Embracing "Enterprise Computing" with Intel servers to replace mainframes will do several things. One, increase the number of software licenses. Two, increase the number of system administrators because there is no uniformity purchasing servers on a project by project basis. Three, expand the size of the data center or construction of new data centers ($$$). Four, drive up network costs (average intel server today requires 4 - 5 connections per OS). I will stop here because these are the major impacts and please don't get me started on the virtualization paradygm. It's not the holy grail for distributed computing.
Your TCA (Total Cost of Acquisition) argument has no merit mostly because acquisition costs are not the 300lb gorilla in the room anymore.
Last, your frequent references to "Legacy" and "modern technology Intel" support that you have not done your homework. I've been hearing the word legacy coined over the past 20 years in reference to the mainframe. If it is so legacy in an industry that has changed so much, why has the technology not been sunset? Counter to that if AMD and Intel are so modern, why have these modern technologies not been embraced on an enterprise level like the mainframe? Do some more research and you will find that AMD and Intel have sought out IBM for improvements and sharing of chip technology designs. Designs that have originated from the mainframe which has some of the most modern technology of any server. Do some more research and you will find the simple concept of "economies of scale" applies very well with the mainframe and fails in a Intel or AMD enterprise deployment model. This concept of TCO, (Total Cost of Ownership) is what will save Europe and the rest of the world real $$$ over the next 20 years in IT spend.