Mainframe Hot Again: Really?
January 21, 2010
I like mainframes. My first content project was completed on one of these svelte—well, maybe that is not the best way to describe—IBM installations. A mainframe when I started fooling around in air conditioned rooms was an complex of hardware, software, cables, and assorted other components. If you wanted to get your class project or a bit of side work done quickly, you learned to work in the “computer center”. None of that waiting in a queue for a key punch machine (anyone remember those?) or for some 17 year old person with thick glasses like moi to put your cards in the processing queue. Nope. Work in the computer center, make some money, and get your work done without meatware latency.
I have written about mainframe in this Web log. Most of the write ups have been in response to IBM public relations and marketing. Now, I surmise, the IBM marketing machine has sent enough signals so that the sensitive antennae of The Economist editorial staff has reacted.
The article that caught my attention is “The Return of the Mainframe: Back in Fashion.” The sub title is, “The mother of all computers no longer looks that old.” The hook for the story is that the Bank of Namibia is getting its first mainframe. Obviously Namibia is proud to be on the bleeding edge of information technology, right out there with Hitachi (er, no), PSI (er, acquired by IBM), T3 (er, in the migration to distributed computing business), and I suppose you can argue that Dell, HP, and Sun, among others are in the “sort of” mainframe business.
I recall reading the 2005 news item that described the mainframe as “a falling star”. If you are not familiar with what’s inside a mainframe installation, check out the specs for various systems on the Tech News site.
The Economist said that mainframes are now distributed systems. I wonder if that means “like Google”? What the Economist told me was:
High “switching costs” explain in large part why mainframes are still a good business for IBM. It is the only big firm left selling them, at prices that start at $100,000 but often reach the millions. Sales of mainframes are said to have brought in about $3.5 billion a year, on average, in the past decade. Although this is only about 3.5% of the firm’s overall revenue, each dollar spent on hardware pulls in at least as much from sales of software and maintenance contracts. Toni Sacconaghi of Bernstein Research estimates that 40% of IBM’s profits are mainframe-related.
Ah, annuity and replacement revenue! I also like “high switching costs”. That is on the money. The reason the Namibia deal is “news” is that I assume it is one of those “emerging markets” that needs a mainframe. Next time I visit Namibia, maybe I will get a chance to see what the country bought?
In my opinion, the killer passage in the Economist story was this one:
More worrying to IBM is a run-in with Neon, a software company. It sells a program that allows computing tasks that usually run on a mainframe’s regular processors to be shifted to the discounted ones meant to run things like Linux. Predictably, IBM is not happy and is said to have threatened to charge higher licensing fees to customers using Neon’s software. This, in turn, has led Neon to file a lawsuit against IBM. Defeat would make a big dent in IBM’s mainframe revenues. Still, the computer industry seems to be moving IBM’s way. The mainframe may well find a new home in corporate computing clouds, the pools of data-processing capacity many firms are building. Many companies are also increasingly interested in buying simpler, more integrated computer systems, even if this means a higher price. Reacting to this, IBM’s rivals are making bets on mainframe-like products. On January 13th HP and Microsoft announced a pact to come up with tight packages of hardware and software. Brad Day of Forrester Research, another market-research group, puts it thus: “We are on the way back to the future.”
Quite a conclusion because IBM has responded to competitors in the mainframe sector by treating them to some of that Big Blue kindness. PSI was bought by IBM. T3 probably has its own view of how IBM has encouraged its business.
The impact of the story on me was:
- IBM is hustling mainframes because it is good money for services
- The marketing muscle of IBM is obvious to me
- The health of the mainframe “industry” is accomplished by shifting the definition of a mainframe to the more fuzzy distributed computing approach
Will I buy a mainframe? Nope. Here’s why:
- IBM hardware is over engineered and designed to create a demand for an IBM trained specialist to perform even basic fixes. FRU means an expensive tech roll and on site charges.
- A mainframe demands an ecosystem of specialists. When that ecosystem is not in place as it is at ArnoldIT.com, then you get to pay folks like ArnoldIT.com to work on mainframe projects. Linux is some help but in other ways, Linux is not the magic wand or part of the magic wand
- There are easier and cheaper ways to accomplish certain computing tasks
- A huge complex called a mainframe that runs Linux is a bit like putting a submarine’s nuclear power generation unit in my garage to generate electricity for my home. Interesting in a theoretical sort of way but otherwise pretty darned crazy.
I think it is super that an emerging market can afford a mainframe, but there may be other ways to provide the computing horsepower required. I am more impressed with the power of the IBM marketing machine.
Stephen E Arnold, January 18, 2010
No one paid me to point out that IBM has marketing muscle. I will report this to the Economic Development Administration at the Department of Commerce.
Comments
3 Responses to “Mainframe Hot Again: Really?”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
[…] MORE […]
The writer of this article is right in many ways. But misses an important point here and there. The statement; “A Mainframe that runs Linux is like…” is true If you run ONE Linux on a Mainframe. But running dozens or even hundreds of Linux’s on a Mainframe is a different thing.
Also today’s Mainframe is not the Mainframe he started his first project on. Mainframe solutions today can (for a large part) be managed by non ArnoldIT people, and this will only get better in the future.
Installing Mainframe products has already become a magnitude easier than ever before (read: http://www.ca.com/files/IndustryAnalystReports/idc__wp_mf_2_initi_sprts_consol_216462.pdf for example). It looks like the discussion always revolves around a “Mainframe VS Distributed” question. I thought we had grown beyond that. Many large enterprises have a mainframe that has a certain (important) role. When new functionality becomes available on the Mainframe, its common business sense to investigate if we can solve existing problems in a different way and if decisions we made in the past may have a different outcome now that the environment has changed.
The #1 issue for large companies is the complexity caused by an unmanageable amount of distributed servers. Distributed Virtualization techniques only solve part of the problem; moving distributed applications to Linux on zOS might solve another part of the problem. I think the IBM Marketing machine could have done a much better job in promoting the mainframe than they have done and many Mainframe clients will agree. Innovation is part of IT and if the environment changes, we should always be open to new possibilities no matter what platform offers them.
Server consolidation to reduce costs and increase reliability is a primary driving force these days, especially in Fortune 500 companies. You may not buy a mainframe but mid and large corporatations will.
IBM System z running Linux is the wave of the future.
“A mainframe demands an ecosystem of specialists.”
A server farm actually requires more attention and costs and specialists? I know – I have worked on both mainframe and server farm platforms.
“A huge complex called a mainframe that runs Linux is a bit like putting a submarine’s nuclear power generation unit in my garage to generate electricity for my home. Interesting in a theoretical sort of way but otherwise pretty darned crazy.”
One IBM System Z mainframe can replace a 1,000 box server farm. Use your common sense on which is significantly less complex and more cost abvantagous.
I am an independent consultant with no corporate affiliations.