I make no secret of the fact that I think the modern mainframe offers an excellent enterprise technology platform. But there are issues—in particular, that mainframe people (and technologies) speak a language all of their own and management is frightened of losing key people (or of losing control of what these people are doing). At one stage, even I had quite a secure job because I knew what things like IEBUPDTE did and could use it for things other people might waste time programming—few others, apparently, could read the IBM manuals.
So, Knowledge Transfer is part of actually exploiting the wonderful mainframe—and tools like CA Mainframe Chorus are an important part of this. And, of course, a special aspect of this issue (particularly in the mainframe area) is the need to transfer knowledge of legacy applications (and their business context) from retiring employees. See, for example, a usefully hard-headed exploration of this issue by David DeLong (a research fellow at the Massachusetts Institute of Technology's Age Lab) here.
However, there's still a place for good old-fashioned training, preferably in an interactive F2F scenario (which helps you learn not only WHAT to do, but WHY; and even WHEN there are special cases and exceptions).
Big mainframe companies like CA Technologies and IBM have always provided mainframe-oriented training and I've been very impressed by the students I've met at, for example, CA Technologies "mainframe university" in Eastern Europe.
Now, CA Technologies has gone a bit further and created an external training program, available to young IT talent outside of CA Technologies, based on its existing internal one. And, just to prove it's serious, it's offering $1M in scholarships for these external students.
If this helps to develop a pool of mainframe expertise independent of any particular vendor, that bodes well for the increased acceptance of the mainframe, especially (I suspect) in emerging economies. Just remember to call it an "Enterprise Server" before the term "mainframe" frightens the horses.
Scholarships will be granted after an evaluation process (using an expert panel which includes representation from the SHARE board of directors), which will look at a candidate's academic achievements, creativity and technology vision. Presumably, candidates will have achieved a prior knowledge of general IT before starting as this is only an 8-week, vendor-agnostic program presenting both real-world and project-based learning—when I was picked up off the streets to become a mainframe DBA many years ago, the Australian Dept. Health put me through a fairly testing 1 year post-grad. sandwich course in Comp. Sci. with practical mainframe training in the vacations.
Still, 8 weeks should be plenty of time for someone with general IT expertise to pick up the specialist aspects of mainframe operations and management. Since I believe that the latest mainframes are a valuable technology resource and that much serious business still runs on legacy mainframe systems, I think that this initiative is a good omen for the future efficiency of enterprise IT.