I do not want people to get the wrong impression, I had some very good times working with the mainframe and I learnt a lot. However, time moves on and I moved with the technology developments to other platforms.

This is very similar to my favourite old car, that was a Saab 9000, 2.3turbo, Anniversary Edition. It had leather seats, wooden fascia, quiet, fast and safe.

I often would still like to have it but for todays requirements it uses too much fuel and would probably start to get too expensive to run. Since that car I have had a Toyota, Saab 9-5 and an Audi. I have choice of the open marketplace, if I stick to the old Saab 9000 I am locked in. Which is never a good thing.

So bottom line, we look at the mainframe with good memories but we need to move forward and adapt to new technologies. We need to let go and adapt. We cannot tow around our old cars with us forever.

The other argument is that that the mainframe is adapting like the Saab 9000 to 9-5, as we get new models. But the new car models do not need to pull or tow along the old models. The problem is that z/OS which comes from ESA, MVS, MVT just keeps on getting add-ons. But it is still has to support all the old apps, mainly based on a batch processing system (which no-one can understand anymore), with some bits to make it run OLTP, Linux and all sorts of goodies (read complexity). Also the CPU's still need to run the S/380, S/390, ESA/390, z/9, z/10 and more recent version of this instruction set, of which there are 100's of old instructions. This is why it is called a CISC processor, Complex Instruction set computer. In comparison to simpler SPARC and Power RISC Reduced Instruction set computers. So we have one hell of a level of complexity with mainframes running complex instructions on RISC processors or derivatives thereof. Now this is interesting and fun for the geeks, but not really for business. As all programs/applications have to go through many levels of instruction abstraction/translation etc. Basically mainframes due to their history have immense complexity. Complexity means costly management and difficult problem determination.

I would have loved to have stayed with the mainframe all my career, it was fun, I was quite good at it, people I worked with were great, but many others also moved to UNIX and open systems. Back in the 1980's I actually believed that I would be working with the mainframe for the rest of my career, but then I was young and there was only IBM and DEC. Chips also changed from Bipolar to CMOS, which made it easier for competition. Now the old heatsink/towers on Bipolar chips look not too dissimilar from all the heatsinks we now need on our PC's sad, that is why Sun coolthreads CPUs are cool. However, technology and the world moved on and the mainframe tries to re-invent itself. While carrying all it's historic baggage, which is some weight that we need to burden ourselves with. So like an old shoe, you can replace the soles so many times until you just have to get a new pair. Sometimes we have to try a new pair of shoes or a new design.