On 10/30/18 1:37 PM, Greg Woods wrote:
On Mon, Oct 29, 2018 at 1:33 PM Rick Stevens <ricks(a)alldigital.com
<mailto:ricks@alldigital.com>> wrote:
Do big IBM (or any) mainframes still exist?
You can still buy S/390's, but the big money is not in mainframes, but
in supercomputers. It used to be (in the days of Seymour Cray) that a
supercomputer just had a really fast processor that could do vector
processing (running the same instruction on a whole block of memory
locations in parallel in a single clock tick), but we have pretty much
gotten to light speed limitations on how fast a processor can be, so
IBM's supercomputers these days are clusters of thousands of processors
and cores, exchanging data over specialized high-speed fabrics such as
Infiniband. Linux is critical to making this work (as well as
specialized application libraries to support interprocess communication
on such as system). The IBM supercomputers I have seen all use a variant
of Red Hat Linux, so I wouldn't be surprised if acquiring Linux
developers might not be the main reason for IBM to want to buy Red Hat.
That's what I was getting at. The original mainframe concept is fairly
passe' now. What we have are massively parallel compute platforms now.
They were very niche before (Cray, Connection Machines, Comprex, et al),
but now seem to be more mainstream and generally more expandable.
I agree that the manner in which Linux (or any truly multi-threaded
system) works makes sense for such platforms. The few supercomputers
I've seen also used Linux or some variant of BSD/Unix. Getting in-house
talent to support those things is a reasonable goal. However, Linux (the
kernel) and Gnu/Linux (the system) have built-in safeguards in their
licensing that would prevent IBM from taking the code private (not that
they wouldn't try).
Now, will they (IBM) allow Red Hat to continue developing (for lack of a
better term) public domain code? That's what worries we Fedorans (sic)
as a group. It's a good question and one we really can't answer at this
time.
I'm hoping IBM leaves Red Hat more-or-less alone. Red Hat's been
successful using their existing model--which is why IBM wanted them in
the first place. There's little good in buggering a working model. If
it's not broken, don't try to fix it. Or, as G. Harry Stine wrote in
"Force of Arms", "If it's stupid and it works, it ain't
stupid!"
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- "I'd explain it to you, but your brain might explode." -
----------------------------------------------------------------------