Two interesting take aways from Martin Fowler’s article on LMAX are: the importance of performance testing, and mechanical sympathy.
Testing is actually one of the most efficient ways of learning modern software systems, so if nothing else it’s going to bring that edge to people. Mechanical sympathy, on the other hand, is one of the enduring qualities of great C/C++ programmers, I’d say that in that case the expression covers a combined affinity with the C++ language standard, the vendor compiler idiosyncrasies, and the particular OS being targeted. I could infer this back in the 1990s as I read books by folks like Herb Sutter, Stan Lippmann, or Scott Meyers. At the time I had a good understanding of the x86 architecture and could see how mechanical sympathy played out – though, I didn’t know the expression until I read Martin Fowler’s blog.
I’ve been focusing on actor based programming models last year or so, I think the LMAX Disruptor is definitely a very interesting and exciting concept.