If you dont give actual sales numbers its normally because you want to present youself stronger than you really are. If the numbers are strong and all curves are going up, its a cheap argument to show. The impression they would like to avoid at all cost is that you bet on a doomed platform, but I'm afraid that by all means the island they are living on is a shrinking one - cost is enormous, development money must be shared between a shrinking user base, know-is fading, ... its not a surprise that none of the younger IT people I know ever wants to work on this machinery. On the other hand I know a lot of (aged, retired) freelancers making good money in keeping vintage systems running.
Yep, just like the naysayers declared the mainframe "dead" and a "dinosaur" back in the early 1990's. The wave of the future was "distributed processing" where data would be available on low end servers and PC's. The industry believed these idiots and it did nearly kill the mainframe market. And what a total bunch of horseshit distributed processing turned out to be. And another factor that nearly killed the mainframe was technology. Up through 3090 mainframe the circuitry was transistor bi-polar. Fast but requires enormous amounts of power and cooling. The next generation machine was already in design when it was apparent that it was going to be a power and cooling monster. It was cancelled in favor of going with CMOS designs. Slower but required much less power and cooling. It took many years to develop that technology to where it is today.
And point of fact...present day....The latest model of Z series is selling like IBM is giving it away. Will it last? Dunno, but I do know there's more to come. So those "experts" who call it a dead legacy and a doomed platform have no clue.
Well, I have worked with IBM mainframes from the beginning of the 90ies (yes, I'm old). I was working for the insurance industry (slow, risk-averse, rich at the times, conservative like hell). There were mainframes and nothing else. IT departments were hierarchical monoliths, the tools for your applications were COBOL, CICS and DB2, and that was it.
We implemented our last mainframe based administration application at the end of 1999. From then on, all was developped on PCs and was multi-platform.
Now make a fast forward towards 2018. *All* of the insurance customers we had have discontinued their mainframe use. The arguments were always the same: IBM drove up the maintenance costs for their legacy licenses up in an incredible way, and other platforms (using Oracle, C++, Java, Web GUIs) were a lot more develop friendly than what IBM tried to sell you as "the future" (z Machines with plugged-on "innovations"). Except for the very big cat companies, their mainframes became a niche market.
Regarding z14, IBM is at the beginning of a new cycle. A look at the graph shows how the other cycles have performed. When new, everything sells well (probably using some good deals), but the long run is what counts. What could be result is that you get the /370 architecture stuff as an add-on to a modern (what CPU ?) high-availability server in future and not vice versa.
I am prepared to bet for a box (20 bottles) of premium Bavarian beer (Augustiner Edelstoff) if one of the following statements is false in 5 years:
- the share of the revenue based on the z Series will be below 7%
- IBM will focus even more on KI, Cloud, ... In order to gain market share, their solutions will be made available on all major platforms, not just z Series.
- z Hardware will be replaced by other emulating CPUs and FPGAs because the shrinking numbers dont justify proprietary chips.
What about it ?