[email protected] (Tony Harminc) writes:
> Which is why I've wondered here why IBM doesn't try to find some
> market for those chips that's different enough from the traditional
> mainframe one that it won't bite into it, but still lets them sell the
> chips for at least something.
>
> Well, maybe they have tried, and maybe there just isn't any such
> market, given the characteristics of the chips: ho-hum
> price/performance, not massively parallel, but extreme on-chip
> reliability. Presumably market segments that need reliability have
> already worked around flaky chips by using other (higher layer)
> approaches.

re:
http://www.garlic.com/~lynn/2014j.html#93 Demonstrating Moore's law

by 30yrs ago, hardware reliability got to the point that majority of
service outages were no longer hardware and shifting to human mistake
and environemtal (power, storms, flooding, fires, etc) ... as a result
next incremental improvement service availability required geographic
replication.

once you have geographic replication for high service availability
... then the replicated systems would also mask any incidental hardware
failure.

in the late 80s and early 90s we did ibm's ha/cmp (high availability)
and we demonstrating superior operational characteristics against pure
hardware fault tolerant. At the time out marketing, I coin'ed the terms
"disaster survivability" and "geographic survivabilty"

Anyway, as a result I got asked to write a section for the ibm corporate
continuous strategy document ... but then it got pulled when both
rochester (as/400) and pok (mainframe) complained they couldn't meet the
objectives.

Then the cluster scaleup part of ha/cmp was transferred, we were told
we couldn't work on anything with more than four processors, and
announced as ibm supercomputer for technical and scientific *ONLY*.
old reference to meeting Ellison's conference room first part
of january1992
http://www.garlic.com/~lynn/95.html#13
within a month of that meeting it had been announced as
ibm supercomputer ... some old email
http://www.garlic.com/~lynn/lhwemail.html#medusa

past ha/cmp posts
http://www.garlic.com/~lynn/subtopic.html#hacmp
past continuous availability posts
http://www.garlic.com/~lynn/submain.html#available

Jim Gray was major person at ibm san jose research creating the original
relational/sql database. When he left for tandem, he palmed a lot of
stuff on me. While at tandem he did a lot of studies&surveys for
availability&outages ... and also the prime mover behind TPC benchmarks.
http://www.tpc.org/information/who/gray.asp
old gray presentation on service outages
http://www.garlic.com/~lynn/grayft84.pdf

-- 
virtualization experience starting Jan1968, online at home since Mar1970

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to