zEC12: I know it's a mainframe, but could we all just stop calling it that? "Enterprise Server" perhaps? Sadly, IBM tells me that the "EC" only implies "Enterprise Class" and they're just using "zEC12" or "zEnterprise EC12" and I suspect that people will mainly associate "z" with mainframes. Sadly, I fear that this may encourage many potential users to think that this is something that doesn't concern them.
But this beast is so much more than the mainframes I knew and loved, good though they were for resilience; security; and high throughput (from using a proper workload scheduler). The danger is, that if you call something a "mainframe" people start assuming that there are still things it doesn't do well, and that it is only really suited for running huge CICS transactional systems in megabanks.
This "mainframe" though, has up to 5.5GHz processors, faster than your desktop PC! And lots of them, in an optimised architecture. And the first support for transactional memory (for improved, more reliable, parallel processing) in a general-purpose commercial machine. And analytics-based management/monitoring. This last is zAware, and it runs in firmware in its own partition, so it can't get the rug pulled from under it when things are going wrong and you really need it. Read more about this in Appendix A in the zEC12 Technical Guide - it's pretty readable and some of the implications of zAware for systems management are pretty cool. Oh, and IBM Technical guides are better than I remember them too - this Appendix even has a neat quote, well worth meditating on: "Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child's? If this were then subjected to an appropriate course of education one would obtain the adult brain". - from Computing Machinery and Intelligence" by Alan Turing.
IBM zSeries has been able to run UNIX and Windows natively (on integrated blade servers) for some time; and it already supports running lots of Linux virtual machines really well. Workload analysis is still an important part of deciding where best to run things but, these days, most workloads run pretty well on z - and some run superbly well.
However, one zEC12 innovation reminds me of the very first mainframe I met: Flash Express SSD memory (Appendix C in the zEC12 Technical Guide, above), intermediate in speed between "real" memory and conventional DASD ("Direct Access Storage Device" - IBM-ese for "hard disk drive"). For now, this supports faster paging - similar to drum memory way-back-when. However, Flash Express will do much more than just that - it already helps to exploit the zEC12's 1MB large pages facility, useful for reducing time collecting data for diagnostic dumps.
What all this really means is that if you need to support secure multi-tenanted Cloud for large numbers of users; or extreme resilience, availability and security; or efficient (approaching 100%) utilisation of your technology with minimised power consumption; and so on; you really should be looking at zEC12, because it has architectural (built-in) features that make these things work reliably, all the time (that is, without "planned downtime"). Assuming, that is, that IBM's pricing models are now, as it claims, more customer-friendly in practice than many old mainframe customers might expect.
I hope to find the time to explore all this, and its implications and any downsides, in more detail, in a research note. Watch this space!