Re: [tlug] Re: Talk about fast HTTP

["Stephen J. Turnbull" <stephen@example.com>]

Bruno Raoult writes:

 > Yes, not your interest, maybe. But you cannot scale for a VAR.

Somebody's getting screwed bad if they're paying to scale value at
risk computations.[1]  GILBO (Garbage In, Lehman Bros. Out ... totally
out<wink>).

Anyway, I'd have to agree with Curt.  Scaling computations is a rather
specialized problem even today.  My brother works for a company whose
product provides better than Infiniband performance (so they say, and
he says it's not marketing hype), and they're surviving, but not
growing explosively yet.  Or maybe it's just that people have no clue;
people *still* pay many man-yen for extra CPU cycles, then starve it
with too-small cache and RAM.

Of course, if you take it a step further, moving data instead of
information is a great way for an IT department to justify a bigger
budget.  Most IT departments don't even do EDP well, of course. :-(

Footnotes: 
[1]  In the back office, surely you're not talking about vector
autoregression.