List of Archived Posts

2007 Newsgroup Postings (05/03 - 05/13)

John W. Backus, 82, Fortran developer, dies

2741s, was Even worse than UNIX

Newbie question on table design

John W. Backus, 82, Fortran developer, dies

Even worse than UNIX

Even worse than UNIX

MTS *FS tape format?

Newbie question on table design

Newbie question on table design

Newbie question on table design

Disc Drives

Newbie question on table design

Newbie question on table design

Interrupts

Newbie question on table design

John W. Backus, 82, Fortran developer, dies

Newbie question on table design

Newbie question on table design

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

John W. Backus, 82, Fortran developer, dies

Newbie question on table design

a little dbms folklore

Even worse than UNIX

Even worse than UNIX

Even worse than UNIX

John W. Backus, 82, Fortran developer, dies

John W. Backus, 82, Fortran developer, dies

IBM Unionization

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

John W. Backus, 82, Fortran developer, dies

Display Technologies Evolution

strange FTP problem--transatlantic asymmetry

Problem with TCP connection close

Newbie question on table design

Disc Drives

z/VM usability

John W. Backus, 82, Fortran developer, dies

z/VM usability

John W. Backus, 82, Fortran developer, dies

IBM Unionization

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

How difficult would it be for a SYSPROG ?

Using rexx to send an email

IBM Unionization

IBM Unionization

John W. Backus, 82, Fortran developer, dies

Using rexx to send an email

John W. Backus, 82, Fortran developer, dies

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

IBM Unionization

John W. Backus, 82, Fortran developer, dies

Lean and Mean: 150,000 U.S. layoffs for IBM?

John W. Backus, 82, Fortran developer, dies

IBM Unionization

Disc Drives

Help settle a job title/role debate

Help settle a job title/role debate

open source voting

Disc Drives

Lean and Mean: 150,000 U.S. layoffs for IBM?

Using rexx to send an email

John W. Backus, 82, Fortran developer, dies

IBM Unionization

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

Linux: The Completely Fair Scheduler

IBM Unionization

IBM 360 Model 20 Questions

IBM 360 Model 20 Questions

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

VLIW pre-history

IBM Unionization

IBM Unionization

John W. Backus, 82, Fortran developer, dies

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

IBM Unionization

VLIW pre-history

VLIW pre-history

VLIW pre-history

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 08:33:33 -0600

Dave Garland <dave.garland@wizinfo.com> writes:

Sarbanes-Oxley?  http://en.wikipedia.org/wiki/Sarbanes-Oxley_Act But
AFAIK that hasn't been eliminated (though businesses would like for it
to be).  That law (stemming from Enron et al.) does things like stiffen
requirements for financial reporting of corporations (and hold
executives liable for fraudulent reports)..

in the financial area there was new qualitative section of basel II
... but that pretty much eliminated before basel II was adopted.

i gave a talk at european finance conference a couple yrs ago that SOX
couldn't prevent/eliminate the ENRON type stuff ... since the criminal
activity was already criminal ... that about the only part of SOX that
was really new was the section at the end on informants.

past posts mentioning basel ii &/or sox
http://www.garlic.com/~lynn/aepay3.htm#riskm The Thread Between Risk Management and Information Security
http://www.garlic.com/~lynn/aadsm5.htm#xmlvch implementations of "XML Voucher: Generic Voucher Language" ?
http://www.garlic.com/~lynn/aepay10.htm#50 glossary
http://www.garlic.com/~lynn/aepay11.htm#29 CIOs Must Be Involved In Controlling Risk In Financial Services
http://www.garlic.com/~lynn/aadsm10.htm#smallpay3 Small/Secure Payment Business Models
http://www.garlic.com/~lynn/aadsm10.htm#cfppki19 CFP: PKI research workshop
http://www.garlic.com/~lynn/aadsm14.htm#50 E-banking is board-level Issue, Says Basel Committee
http://www.garlic.com/~lynn/aadsm14.htm#52 Committee calls for better e-banking security management
http://www.garlic.com/~lynn/aadsm16.htm#7 The Digital Insider: Backdoor Trojans ... fyi
http://www.garlic.com/~lynn/aadsm19.htm#10 Security as a "Consumer Choice" model or as a sales (SANS) model?
http://www.garlic.com/~lynn/aadsm21.htm#3 Is there any future for smartcards?
http://www.garlic.com/~lynn/aadsm22.htm#26 FraudWatch - Chip&Pin, a new tenner (USD10)
http://www.garlic.com/~lynn/aadsm23.htm#10 PGP "master keys"
http://www.garlic.com/~lynn/aadsm25.htm#12 Sarbanes-Oxley is what you get when you don't do FC
http://www.garlic.com/~lynn/aadsm25.htm#13 Sarbanes-Oxley is what you get when you don't do FC
http://www.garlic.com/~lynn/aadsm25.htm#14 Sarbanes-Oxley is what you get when you don't do FC
http://www.garlic.com/~lynn/aadsm25.htm#15 Sarbanes-Oxley is what you get when you don't do FC
http://www.garlic.com/~lynn/aadsm25.htm#26 Fraudwatch - how much a Brit costs, how to be a 419-er, Sarbanes-Oxley rises as fraud rises, the real Piracy
http://www.garlic.com/~lynn/aadsm25.htm#43 Audit Follies - Atlantic differences, branding UnTrust, thunbs on Sarbanes-Oxley, alternates
http://www.garlic.com/~lynn/aadsm26.htm#2 Audit Follies - Atlantic differences, branding UnTrust, thunbs on Sarbanes-Oxley, alternates
http://www.garlic.com/~lynn/2003k.html#41 An Understanding Database Theory
http://www.garlic.com/~lynn/2005k.html#23 More on garbage
http://www.garlic.com/~lynn/2005t.html#26 Dangerous Hardware
http://www.garlic.com/~lynn/2006h.html#33 The Pankian Metaphor
http://www.garlic.com/~lynn/2006h.html#58 Sarbanes-Oxley
http://www.garlic.com/~lynn/2006i.html#1 Sarbanes-Oxley
http://www.garlic.com/~lynn/2006j.html#28 Password Complexity
http://www.garlic.com/~lynn/2006o.html#35 the personal data theft pandemic continues
http://www.garlic.com/~lynn/2006u.html#22 AOS: The next big thing in data storage
http://www.garlic.com/~lynn/2007b.html#63 Is Silicon Valley strangeled by SOX?

2741s, was Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: 2741s, was Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 09:44:47 -0600

johnl@iecc.com (John L) writes:

The only problem with 2741s was that they were extremely unreliable.
The Selectric mechanism wasn't intended to be driven at full speed
page after page. In the terminal room at Princeton I used, I don't
ever remember all of the dozen or so 2741s working at the same time.
KSR 33's, on the other hand, just kept on working.

depends on the duty cycle they were manufactured for ... 1052-7
operator's consoles used essentially same golf ball mechanism ... but
were manufactured for significantly heavier duty cycle.  in fact most of
the 1052 models were built for significantly heavier duty than 2741s.

i had 2741s for personal use (one at office and one at home for several
yrs w/o problems).

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Thu, 03 May 2007 10:10:10 -0600

"David Cressey" <cressey73@verizon.net> writes:

I know practically nothing of CDC culture, but quite a bit about DEC
culture, going way back.  My impression of CDC culture, gleaned indirectly
from what Niklaus Wirth had to say about the CDCmachines,  is that CDC
culture discovered interactive development later than DEC culture did.  I'm
just about certain that IBM culture discovered interactive development later
than DEC culture did.  This is somewhat related to the topic at hand.

I've often commented that it wasn't that IBM culture didn't have
interactive development ... which was compareable to features/size of
most other vendors (that might be considered interactive) ... it was
that in the 60s, 70s & much of the 80s, the batch market size
dwarfed the interactive.

in afc ng ... i've often mentioned that the number of vm/43xx customer
installs were larger than vax/vms installs ... in part because some
number of large customers would order then in blocks of 100s at a time
(i don't believer there were ever any single vax/vms orders for 1000
machines ... until possibly you got to microvax). misc. old email
discussing 43xx activity
http://www.garlic.com/~lynn/lhwemail.html#43xx

internally ... all the original relational/sql work was done on vm ...
lots of past system/r posts
http://www.garlic.com/~lynn/submain.html#systemr

and the internal network was larger than the (whole) internet/arpanet
from just about the beginning until sometime mid-85 ... and was nearly
all vm machines. misc. past internal network postings
http://www.garlic.com/~lynn/subnetwork.html#internalnet

one of my hobbies was building and supporting highly modified custom
operating system for internal distribution. i've periodically joked that
the number of customer batch systems were much larger than the number of
customer vm systems. the number of customer vm systems were much larger
than the total number of internal vm systems. the total number of
internal vm systems were much larger than the peak number of internal vm
systems that i provided distribution and support for. however, that peak
number (that i directly built, distributed, etc) was still as large as
the total number of multics systems that ever existed. this comparison
was somewhat because the vm stuff had started on the 4th flr of 545 tech
sq ... and the multics stuff was on the 5th flr ... misc. past posts
about science center and 545 tech sq
http://www.garlic.com/~lynn/subtopic.html#545tech

in the early 80s ... bitnet/earn for a time was compareable in size
to internet. bitnet/earn was based on similar vm technology used
in the internal network ... but totally different network
http://www.garlic.com/~lynn/subnetwork.html#bitnet

it wasn't that the IBM culture involved in interactive was smaller
than other vendors interactive activity ... it was that the batch
market acitivty was so much larger, that skewed perception.

for some topic drift ... post mentioning recent news article
http://www.garlic.com/~lynn/2007i.html#68 A tribute to Jim Gray

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 11:12:48 -0600

Charlton Wilbur <cwilbur@chromatico.net> writes:

No, I'm assuming that virtual payments mean I will need to instruct
some entity to transfer money from one account to another.  Some
identifying information will need to be made public, or how will I
tell the money where to go?

Either I need to provide my account number to the person I want to
give money to so that they can withdraw the money, or the person I
want to give money to needs to give me his or her account number so
that I can send the money.  Alternately, we could both give our
account numbers to a trusted third party.  But you can't get around
needing to give out the account number or other personally identifying
information.

re:
http://www.garlic.com/~lynn/2007i.html#65 John W. Backus, 82, Fortran developer, dies

the issue isn't so much that you divulge the account number ... it is
that the account number (and other "static" information) is sufficient
to authenticate and perform a transactions .... making the
infrastructure vulnerable to replay attacks.

some recent posts about authentication independent of the transaction
and authentication tied to the transaction (i.e. when they are separate,
it is possible that the infrastructure becomes vulnerable to other
kinds of attacks ... like man-in-the-middle attacks):
http://www.garlic.com/~lynn/2007i.html#73 public key pas>sword authentication
http://www.garlic.com/~lynn/2007i.html#74 public key password authentication
http://www.garlic.com/~lynn/aadsm26.htm#65 Public key encrypt-then-sign or sign-then-encrypt?
http://www.garlic.com/~lynn/aadsm26.htm#66 More Tipping Point evidence -POS vendors sued

in the past, i've mentioned that one-time account numbers (good for one
use only) have been tried as countermeasure to replay attacks:
http://www.garlic.com/~lynn/2007h.html#36 sizeof() was: The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007h.html#43 sizeof() was: The Perfect Computer - 36 bits?

this week ... there was announcement of a different kind of one-time
value (as countermeasure to replay attacks):

VeriSign to use one-time passwords for bank cards
http://www.networkworld.com/news/2007/050207-verisign-to-use-one-time-passwords.html
VeriSign looks to offer bank cards with one-time passwords
http://www.computerworld.com/action/article.do?command=viewArticleBasic&articleId=9018518
VeriSign will ship two-factor authentication for debit cards
http://www.theregister.co.uk/2007/05/01/verisign_twofactor_authentication/
VeriSign to offer passwords on bank card
http://www.usatoday.com/tech/news/computersecurity/2007-05-01-verisign-passwords_N.htm?csp=34

note that just having unique per transaction information ... aka or
"dynamic data" ... while it may be countermeasure to replay attacks
... it doesn't necessarily mean that it isn't vulnerable to things like
man-in-the-middle attacks ... it depends on details of the
implemenation. for example, in the posts about yes cards
... "dynamic data" is suggested as countermeasure to the yes
card replay attack vulnerability. However, the "dynamic
data" implementation may still be vulnerable to man-in-the-middle
attacks
http://www.garlic.com/~lynn/subintegrity.html#yescard

in posts about the "naked transaction" metaphor ... the countermeasure
integrates the authentication and integrity operations with the actual
transaction ... in effect "armoring" the transaction ... as
countermeasure to wide range of attacks (including replay
attacks and man-in-the-middle attacks)
http://www.garlic.com/~lynn/subintegrity.html#payments

and misc. posts mentioning man-in-the-middle attacks
http://www.garlic.com/~lynn/subintegrity.html#mitm

now, there may some additional confusion about two-factor
authentication ... referenced in one of the articles.

nominally, 3-factor authentication refers to
http://www.garlic.com/~lynn/subintegrity.html#3factor

• something you have
• something you know
• something you are

... now, normally, multi-factor authentication is assumed to be more
secure because the different factors are subject to different threats
and exploits ... for example PINs (i.e. something you know) are
considered countermeasure to lost/stolen cards (i.e. something you
have) ... modulo writting the PIN on the card.

dynamic data built into the card ... isn't considered multi-factor
authentication ... i.e. providing different countermeasures
to a single threat ... like lost/stolen card. dynamic data built into
the card is more secure form of a single mode of authentication
... providing countermeasure to replay attacks ... but
can still vulnerable to single threat from a lost/stolen card.

Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 11:48:59 -0600

timcaffrey@aol.com (Tim McCaffrey) writes:

Of course, with IDE drives looking to move to 4K sectors, the side-effect of a
single sector failure for databases can be all that more annoying.

all other things being equal, any specific 4k record might have 8times
the probability of failing as 512 byte record (being 8times as large
... also comes up in things like yield when manufacturing large sized
chips).

however, the better FEC may reduce the overall number of record hard
failures that a disk might have i.e. improves the overall reliability
of a drive while increasing the probability of any specific record
failure by something less than 8times. raid has become much more
common place as countermeasure to both drive and record
failures.

in the 80s ... when we were doing HSDT project
http://www.garlic.com/~lynn/subnetwork.html#hsdt

we were working with a company called cyclotomics that specialized in
reed-solomon error correcting codes ... for some of our high-speed
links.  cyclotomics had also been active in FEC for cdrom standard (with
2k byte blocks) ... and were eventually bought by Kodak.

some of this was because some of the links where high-speed satellite
links (i.e. on a link getting 10**-9 bit-error-rate, 15/16ths
reed-solomon FEC could provide the effective equivalent of 10**-15
bit-error-rate ... i.e. about six orders of magnitude improvement). we
also had somebody that was considered one of the five best satellite
RF engineers in the world and who also happened to be a graduate
student of reed's at caltech. but then ... he would also say that his
favorite instructor at mit in his undergraduate days had been my
wife's father.

in any case, i was trying to really push the envolope (at the time)
... trying to do an interface that would sustain up to 3mbytes/sec
... do reed-solomon FEC and do crypto (along with being able to handle
changing crypto key on every packet ... i.e. stepup from raw link
encryptor).

past posts mentioning reed-solomon and/or other FEC work:
http://www.garlic.com/~lynn/93.html#28 Log Structured filesystems -- think twice
http://www.garlic.com/~lynn/99.html#115 What is the use of OSI Reference Model?
http://www.garlic.com/~lynn/99.html#210 AES cyphers leak information like sieves
http://www.garlic.com/~lynn/2000c.html#38 Does the word "mainframe" still have a meaning?
http://www.garlic.com/~lynn/2001.html#1 4M pages are a bad idea (was Re: AMD 64bit Hammer CPU and VM)
http://www.garlic.com/~lynn/2001b.html#80 Disks size growing while disk count shrinking = bad performance
http://www.garlic.com/~lynn/2001k.html#71 Encryption + Error Correction
http://www.garlic.com/~lynn/2002p.html#53 Free Desktop Cyber emulation on PC before Christmas
http://www.garlic.com/~lynn/2003e.html#27 shirts
http://www.garlic.com/~lynn/2003h.html#3 Calculations involing very large decimals
http://www.garlic.com/~lynn/2003j.html#73 1950s AT&T/IBM lack of collaboration?
http://www.garlic.com/~lynn/2004f.html#37 Why doesn't Infiniband supports RDMA multicast
http://www.garlic.com/~lynn/2004h.html#11 Mainframes (etc.)
http://www.garlic.com/~lynn/2004o.html#43 360 longevity, was RISCs too close to hardware?
http://www.garlic.com/~lynn/2005k.html#25 The 8008
http://www.garlic.com/~lynn/2005n.html#27 Data communications over telegraph circuits
http://www.garlic.com/~lynn/2005r.html#52 Go-Back-N protocol?
http://www.garlic.com/~lynn/2005t.html#50 non ECC
http://www.garlic.com/~lynn/2006u.html#44 waiting for acknowledgements
http://www.garlic.com/~lynn/2006u.html#45 waiting for acknowledgements
http://www.garlic.com/~lynn/2007.html#29 Just another example of mainframe costs

Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 15:56:43 -0600

"David Wade" <g8mqw@yahoo.com> writes:

Linux already will do record orientated I/O thats why C has two I/O
paradigms.Looking for an <LF> character seems pretty trivial compared to the
overhead of MVS variable length i/o which has logical ecord sizes on the
front of each record. Even worse if you are doing variable block size on
tape where you have the block size in the front as well. Also note that MVS
disks (usually known as DASD) can have each track formatted with a different
block size. In practice a given file will usually have the same block size
for all blocks, so for cards sized records they will often be blocked 800 or
8000, whereas for a file with longer records bigger blocks will be used.
Thats why you need tables like this to "allocate" a file:-

http://sdisw.com/vm/dasd_capacity.html

Thank goodness for fixed block disks...

vm & cms used logical fixed-block from the start ... even on ckd
disks. normal os/360 genre would dynamically physically format dasd
track. vm/cms would initialize dasd tracks to standard format.

these days i believe all "DASD" is physically some form of fixed-block
... with controller providing the magic emulation of CKD.

x-over post with old formulas for calculating overhead number of records
per physical track ... for various tracks and tecnologies
http://www.garlic.com/~lynn/2006m.html#5 Track capacity?
http://www.garlic.com/~lynn/2006m.html#8 Track capacity?
http://www.garlic.com/~lynn/2006m.html#13 Track capacity?
http://www.garlic.com/~lynn/2006v.html#31 MB to Cyl Conversion

... i.e. from q&d conversion of old gcard ios3270 to html
http://www.garlic.com/~lynn/gcard.html#26.3

MTS *FS tape format?

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: MTS *FS tape format?
Newsgroups: alt.folklore.computers
Date: Thu, 03 May 2007 16:08:59 -0600

"David Wade" <g8mqw@yahoo.com> writes:

I only ever used MTS on the pre 3270 (I forget the model) displays. I seem
to recall that the screen was buffered in controller memory. How I wish I
had kept the manuals from then...

2260? ... here is mts reference mentioning 2260
http://www.eecis.udel.edu/~mills/gallery/gallery8.html

above also mentions reference to 2703 terminal controller and designing
something better from pdp with  pointer to here:
http://www.eecis.udel.edu/~mills/gallery/gallery7.html

when i was an undergraduate ... i was involved in doing something
similar ... but building a "better" terminal controller out of
interdata/3 (which evolved into combination of interdata/4 and
interdata/3s). there was some subsequent article about getting blameed
for contributing to the clone controller business .... lots of past
posts
http://www.garlic.com/~lynn/subtopic.html#360pcm

later interdata was bought and the boxes marketed under the perkin/elmer
brand. within the past decade ... i even ran into installation with
still operating perkin/elmer box handling a lot of traffic in a large
transaction processing datacenter.

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Thu, 03 May 2007 17:58:42 -0600

paul c <toledobythesea@oohay.ac> writes:

also, not sure if you could even buy them then or around 1965 when the
360's made their splash, the price was nominal to satisfy accountants,
either or both corporate or governmental and leasing was the only way
to acquire.  the even bigger money was in maintenance fees and buyer
lock-in.

much smaller world then.  around 1990, i asked some knowledgeable
hardware types how many mainframes there were in the world and how
many were sold per year, of 3033 class (speed of what was called a
"couple of mips") and the general concensus seemed to be several
thousand, with a big sales year for Hitachi, Amdahl and IBM consisting
of several hundred boxes sold.  i know some 4341/4381's approached
that mip rating, but find it hard to believe that there were many
customers who ordered hundreds of them.

Multiply those numbers by a few dozen or a hundred real programmers
for each cpu and it's not hard to see why brain-washed oldtimers wax
nostalgic or why there is so much commercial chaos now, with hundreds
of thousands of accomplished programmers who have an equal right to
believe they have seen the truth.

re:
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design.

4341 was around mip (early models slightly less, later ones more).

both vax/vms and 43xx machines appeared to drop below threshhold and
started trend towards department machines. by the mid-80s and 4381
time-frame ... workstations and larger PCs were started to take over
that market segment.

posting
http://www.garlic.com/~lynn/2001m.html#15 departmental servers
with old email reference to a customer upgrading/changing order from 20
to a couple hundred over a period of approx. six months.
http://www.garlic.com/~lynn/2001m.html#email790404b

internally the explosion in the machines as departmental machines had
a large number of the conference rooms (like in STL and other
locations) being taken over for 43xx departmental machines

major market segment that were early into ordering hundreds at a time
was chip industry (for running chip design tools ... things that
would eventually migrate to the emerging high-end workstations).

old post with a decade of vax/vms machines sliced & diced by model,
year, us/non-us
http://www.garlic.com/~lynn/2002f.html#0 Computers in Science Fiction

43xx ... aka 4331 (follow-on to 135/138) and 4341 (follow-on to
145/148) shipped more in that mid-range market than DEC. both 43xx
follow-on and dec vax/vms show the effect of workstations and large
PCs starting to take over that mid-range market in the mid-80s.

six 4341s would provide higher aggregate performance than 3033 at
lower cost ... and were easier to justify and easier to house.  some
of the old 43xx email
http://www.garlic.com/~lynn/lhwemail.html#43xx

includes discussions working with the disk division on moving various
disk design and development tools to clusters of 43xx machines
... that resided in various locations outside the datacenter (and off
of the large mainframes that required extensive faciilties, cooling,
and space requirements in large datacenters). part of the discussion
was that there was never going to be any way that they were going to
be able to cost justify and/or provide physical facilities ... for the
amount of processing that was going to be needed (assuming the
high-end mainframes).

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Thu, 03 May 2007 18:10:41 -0600

paul c <toledobythesea@oohay.ac> writes:

Lynn, Thanks, have you any sales numbers?

re:
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design.

old post with a decade of vax/vms machines sliced & diced by model,
year, us/non-us
http://www.garlic.com/~lynn/2002f.html#0 Computers in Science Fiction

i.e. a decade of vax/vms machines "sales", sliced & diced by model,
year, us/non-us

and 43xx shipped more to customers than vax/vms.

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Thu, 03 May 2007 20:14:20 -0600

paul c <toledobythesea@oohay.ac> writes:

coming back now, i think it was the 2321.  with any luck, i'll blow a
valve before everybody else gets sick of this tripe.

re:
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design.
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design.
http://www.garlic.com/~lynn/2007j.html#8 Newbie question on table design.

x-over thread from mainframe newsgroup mentioning 2321:
http://www.garlic.com/~lynn/2007e.html#38 FBA rant
http://www.garlic.com/~lynn/2007e.html#51 FBA rant
http://www.garlic.com/~lynn/2007e.html#63 FBA rant
http://www.garlic.com/~lynn/2007e.html#64 FBA rant
http://www.garlic.com/~lynn/2007f.html#0 FBA rant
http://www.garlic.com/~lynn/2007f.html#3 FBA rant
http://www.garlic.com/~lynn/2007f.html#5 FBA rant
http://www.garlic.com/~lynn/2007f.html#12 FBA rant

Disc Drives

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Disc Drives
Newsgroups: alt.folklore.computers
Date: Fri, 04 May 2007 06:39:07 -0600

isw <isw@witzend.com> writes:

Drive manufacturers often quote something like a "five year" service
lifetime for drives. I am led to believe that it's due to bearing
failure. Is that not correct?

recent post
http://www.garlic.com/~lynn/2007h.html#13 Question on DASD Hardware

with a few references to report published by google on detailed
statistics on disk drive failures

and a couple other references ...

Hard-Drive Failures Surprisingly Frequent
http://www.pcworld.com/article/id,131168-c,harddrives/article.html

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Fri, 04 May 2007 06:57:10 -0600

"David Cressey" <cressey73@verizon.net> writes:

Fair enough.  IBM culture was big enough, at the time, so it could
accommodate a large number of internal subcultures.  The part of IBM culture
that was visible to me was definitely not into interactive development.

Even though they had interactive terminals, on line editing, etc. etc.
compiling a source program was a batch job.  And that affected the workflow.

re:
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#8 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#9 Newbie question on table design

no, the comment was the "customer" market size for interactive
computing was compareable to other vendor's interactive computing
market size.

that is separate from the comment that the internal interactive use
was extensive.

however, the observation was that the batch market size was so much
larger (than either) ... that it skewed a lot of perception.

there were several internal battles over feature/function of 3270
terminals ... with product managers frequently claiming the 3270 major
market was for "data entry" (related to batch environments) as opposed
to interactive computing (in part ... again ... because the "data
entry" market size was so much larger than the interactive computing
market size).

for some topic drift ... part of it was resolved with the introduction
of ibm/pc. i've frequently commented that a big part of the uptake for
ibm/pc was that there was large business volume in 3270 desktop
terminals. ibm/pc cost about the same as 3270 terminal ... and it
could emulate a 3270 terminal with some local application software
capability ... in a single desktop footprint. it would be an easy
business justification no-brainer to switch the 3270 terminal
allocated budget from real 3270s to ibm/pc. lots of past posts related
to terminal emulation and market uptake for ibm/pc
http://www.garlic.com/~lynn/subnetwork.html#emulation

the ("original") virtual machine system was cp40 developed at the
science center
http://www.garlic.com/~lynn/subtopic.html#545tech

on a specially modified 360/40 with virtual memory hardware ...  and
"CMS" (cambridge monitor system) for interactive computing. When
standard 360/67 with virtual memory became available, cp40 was ported
and renamed cp67. later with the introduction of virtual memory
standard on all 370s, cp67 morphed into vm370 (and cms was renamed to
the conversational monitor system).

the really big volumes in vm370 started to appear with 4341s ...  with
the explosion in the mid-range market segment (also seen by DEC with
vax/vms). subsequently in the mid-80s, this market segment started
moving to workstations and large PCs.

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Fri, 04 May 2007 07:16:04 -0600

Anne & Lynn Wheeler <lynn@garlic.com> writes:

on a specially modified 360/40 with virtual memory hardware ...  and
"CMS" (cambridge monitor system) for interactive computing. When
standard 360/67 with virtual memory became available, cp40 was ported
and renamed cp67. later with the introduction of virtual memory standard
on all 370s, cp67 morphed into vm370 (and cms was renamed to the
conversational monitor system).

the really big volumes in vm370 started to appear with 4341s ...  with
the explosion in the mid-range market segment (also seen by DEC with
vax/vms). subsequently in the mid-80s, this market segment started
moving to workstations and large PCs.

re:
http://www.garlic.com/~lynn/2007j.html#11 Newbie question on table design

for other topic drift ... the original relational/sql implementation was
done at SJR on vm370 (in the late 70s, backus office was about six doors
down from mine, and codd's office was upstairs from mine). lots
of past references to system/r
http://www.garlic.com/~lynn/submain.html#systemr

then there was system/r technology transfer from SJR to Endicott for
SQL/DS (i.e. vm370, vs/e, 4341s, etc).

for some other drift ... one of the people in the meeting mentioned
here
http://www.garlic.com/~lynn/95.html#13
http://www.garlic.com/~lynn/96.html#15

claimed to have handled majority of the technology transfer from
Endicott back to STL for DB2 (even tho SJR and STL were only about 10
miles apart). This was all "mainframe" and mostly implemented in PLS.
(DB2 on the corporation's mainframe "batch" platform, MVS)

Later there was "SHELBY" project done in Toronto to do a relational
DBMS implementation for OS2 implemented in C. This became available on
a number of (non-mainframe) platforms and is also marketed as DB2.  a
few past posts mentioning shelby:
http://www.garlic.com/~lynn/2005b.html#1 Foreign key in Oracle Sql
http://www.garlic.com/~lynn/2005u.html#41 Mainframe Applications and Records Keeping?
http://www.garlic.com/~lynn/2006w.html#13 IBM sues maker of Intel-based Mainframe clones

Interrupts

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Interrupts
Newsgroups: alt.folklore.computers
Date: Fri, 04 May 2007 09:24:17 -0600

jmfbahciv writes:

And the Computer Center has a Cray for you.  It has never failed
before when you've used it for your long runtime calculations.
Are you, the researcher, going to spend your valuable time
writing checkpoint software or are you going to work on the
numerical analysis code?

I know what the chocie will be...and it ain't checkpointing
for a system that has never been known to crash.

in much of the 70s, SJR ran MVT 370/195 service ... that a lot of
organizations used, however, backlog & turn around could be measured in
months.

one of the things i've mentioned before is when bldg. 15 (disk product
test lab) got 3033 ... we were able to move air bearing simulation
application (part of designing floating thin-film heads) move the 195
across the street to the 3033 ... getting significantly better
turn-around ... a couple recent posts
http://www.garlic.com/~lynn/2007f.html#46 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007i.html#83 Disc Drives

the 3033 had about half the thruput of the peak 195 thruput ... however
a 3033 with effectively zero percent utilized would have much better
turn-around than waiting weeks for 195 turn around.

palo alto science center also was getting lengthy turn-arounds on some
of their numerical intensive workload being run by sjr's 370/195. they
had a 370/145 running vm (they had done both the morph from cms\apl to
apl\cms as well as the apl microcode assist on the machine) ... which
had about 1/30th the thruput of 195. The 370/195 batch workload ran
24hrs/day 7day/week. Palo Alto's vm interactive service was primarily
first shift and the machine was mostly unloaded 2nd, 3rd and
weekends. Palo Alto did some checkpoint support and started the
application on their 145 in the background ...  so it got whatever
spare cycles were available 1st shift and much of the off-shift
processing cycles. While it might take the application several weeks
to complete on the 145 ... that was better turn-around than they were
getting from the sjr 195.

misc. past posts mentioning palo alto moving numerical intensive
application from sjr 195 to their own 145 (that only had about
1/30 the processing rate)
http://www.garlic.com/~lynn/2001g.html#22 Golden Era of Compilers
http://www.garlic.com/~lynn/2001n.html#39 195 was: Computer Typesetting Was: Movies with source code
http://www.garlic.com/~lynn/2002n.html#63 Help me find pics of a UNIVAC please
http://www.garlic.com/~lynn/2002q.html#27 Beyond 8+3
http://www.garlic.com/~lynn/2003c.html#77 COMTEN- IBM networking boxes
http://www.garlic.com/~lynn/2003d.html#54 Filesystems
http://www.garlic.com/~lynn/2003j.html#69 Multics Concepts For the Contemporary Computing World
http://www.garlic.com/~lynn/2004.html#21 40th anniversary of IBM System/360 on 7 Apr 2004
http://www.garlic.com/~lynn/2004k.html#34 August 23, 1957
http://www.garlic.com/~lynn/2005f.html#4 System/360; Hardwired vs. Microcoded
http://www.garlic.com/~lynn/2005o.html#44 Intel engineer discusses their dual-core design
http://www.garlic.com/~lynn/2005s.html#34 Power5 and Cell, new issue of IBM Journal of R&D
http://www.garlic.com/~lynn/2006b.html#8 Free to good home: IBM RT UNIX
http://www.garlic.com/~lynn/2006c.html#44 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006d.html#0 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006d.html#4 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006d.html#16 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006x.html#27 The Future of CPUs: What's After Multi-Core?

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Fri, 04 May 2007 09:59:58 -0600

"David Cressey" <cressey73@verizon.net> writes:

You are absolutely right.  I can back that up from my own experience.  At
about the 1969 time frame,  I had been using interactive timesharing systems
for a while,  and I knew how to use the debugger and the text editor
amazingly well.  But my ability to design   a piece of software that was
logically correct (barring minor errors) at the outset was, in retrospect,
less than it might otherwise have been.

Between 1969 and 1980, due to a series of job changes and outlook changes,
I migrated from assembler to Pascal.  Pascal was the first language that I
learned thoroughly before writing any programs.  And, in the Pascal learning
materials,  there was a great deal about designing a program correctly.

Wirth had the attitude that there was no reason why a carefully written
program shouldn't compile on the first try.   I never got as rigid as that.
But Pascal ended up being the main reason I never learned the VAX debugger.
The VAX debugger was, by all accounts,  much more powerful than the earlier
more primitive tools I had used  (DDT for example).

re:
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#8 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#9 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#11 Newbie question on table design

for other topic drift ... mention that system/r, sql/ds and (mainframe)
DB2 was primarily PLS
http://www.garlic.com/~lynn/2007j.html#12 Newbie question on table design

slight drift with some past comments about PLS in relationship to
System/R implementation
http://www.garlic.com/~lynn/2007d.html#17 Jim Gray Is Missing

however, i've mentioned also working on a "relational" DBMS
implementation for the Los Gatos VLSI tool group (that had some
participation from some people in STL). This was relational in the
sense that all the relations were directly instantiated ... rather
than a data dictionary that applied to a uniform table structure. It
shared characteristics with System/R that the "relations" were
abstracted with indexes under the covers (and not exposed as part of
the data).

The tools group had been doing extensive work on various kinds of
languages using Metaware's TWS ... misc. past refs:
http://www.garlic.com/~lynn/2005e.html#0 [Lit.] Buffer overruns
http://www.garlic.com/~lynn/2005e.html#1 [Lit.] Buffer overruns

and developed a Pascal implementation that was used extensively for
numerous tools ... including the (tools group) DBMS implementation
... recent reference:
http://www.garlic.com/~lynn/2007h.html#41 Fast and Safe C Strings: User friendly C macros to Declare and use C Strings
http://www.garlic.com/~lynn/2007h.html#61 Fast and Safe C Strings: User friendly C macros to Declare and use C Strings

it was also eventually released as vs/pascal language product.

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Fri, 04 May 2007 10:28:08 -0600

Anne & Lynn Wheeler <lynn@garlic.com> writes:

Study: Signature Debit Fraud Runs 15 Times Higher Than on PIN Debit
http://www.digitaltransactions.net/newsstory.cfm?newsid=738

re:
http://www.garlic.com/~lynn/2007i.html#58 John W. Backus, 82, Fortran developer, dies
http://www.garlic.com/~lynn/2007i.html#59 John W. Backus, 82, Fortran developer, dies

slightly related article:

The T.J. Maxx Hack Illuminates Reality Credit Card Companies Want to
Hide from Online Merchants, Says Payment Service Provider
http://www.prweb.com/releases/2007/5/prweb522888.htm

and old post on security proportional to risk
http://www.garlic.com/~lynn/2001h.html#61

and reference to attackers frequently are easily able to
outspend defenders:
http://www.garlic.com/~lynn/2007e.html#26 Securing financial transactions a high priority for 2007
http://www.garlic.com/~lynn/2007g.html#20 T.J. Maxx data theft worse than first reported
http://www.garlic.com/~lynn/2007h.html#56 T.J. Maxx data theft worse than first reported
http://www.garlic.com/~lynn/2007i.html#64 John W. Backus, 82, Fortran developer, dies

lots of general posts mentioning fraud, risks, threats, vulnerabilities,
and/or exploits
http://www.garlic.com/~lynn/subintegrity.html#fraud

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Fri, 04 May 2007 12:13:01 -0600

paul c <toledobythesea@oohay.ac> writes:

Nit: I think this should have said "typical" IBM culture.  I once
worked in an atypical IBM shop where, for example, very few
programmers knew any JCL, that included assembler programmers.  Nearly
everybody used TSO and TSOTEST exclusively for compiles and
assemblies, so-called "link-edits" and debugging.  There was no
waiting for batch jobs to start.  Typical IBM installations usually
forbade this for cost reasons but a few bought enough hardware to
allow it.  I found it just as productive as using the early Borland
"Turbo" integrated environments on a dedicated PC.  Very few people
needed to understand how to write tso "clist" scripts, just as very
few Borland customers needed to modify the standard "workflow".  One
could expect several thousand lines to assemble and link with other
code in as little as ten seconds and usually compiles too.  One
typically was dealing with several hundred thousand lines in any given
component, but things were arranged so that usually only a handful of
modules needed to re-compiled.

for much of the seventies, SJR ran batch MVT service on 370/195. There
was some numerical intensive "JOBs" that could have several week turn
around. recent post with reference to palo alto science center moving
one of their jobs to background on their vm/145 (still talking several
weeks to complete ... but turn around was shorter/less than what they
were getting on the 195 ... peak performance of 195 was about 30times
that of 145):
http://www.garlic.com/~lynn/2007j.html#13 Interrupts

with respect to pascal comments in previous post:
http://www.garlic.com/~lynn/2007j.html#14 Newbie question on table design.

this post
http://www.garlic.com/~lynn/2007h.html#41 Fast and Safe C Strings: User friendly C macros to Declare and use C Strings

mentions much later when the corporation had decided to transition to
some number of COTS VLSI tools ... as part of that transition they were
providing some of the VLSI tool vendors copies of some of the internal
applications.

by that time, vs/pascal was supported for both mainframe operation and
rs/6000 (aix) operation. however, as part of the COTS tools transition,
various of internal tools had to also run on other vendor workstations.
I had opportunity to port one 60,000 line pascal application to other
platforms ... however, it seemed like these other implementations had
never gotten much past the student programming application stage (and
never dealt with a 60,000 line application).

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Fri, 04 May 2007 12:49:39 -0600

"David Cressey" <cressey73@verizon.net> writes:

Fair enough.  IBM culture was big enough, at the time, so it could
accommodate a large number of internal subcultures.  The part of IBM culture
that was visible to me was definitely not into interactive development.

Even though they had interactive terminals, on line editing, etc. etc.
compiling a source program was a batch job.  And that affected the workflow.

re:
http://www.garlic.com/~lynn/2007j.html#11 Re: Newbie question on table design.
http://www.garlic.com/~lynn/2007j.html#12 Re: Newbie question on table design.

one of the things you started to see with cp/cms in the 60s was the
appearance of a number of personal computing applications (similar to
various of the things that you would later see on PCs. in the 80s).

in the 60s, there were two operations that sort of spun off of the
science center
http://www.garlic.com/~lynn/subtopic.html#545tech

to offer commercial time-sharing services. they transitioned to 370
machines when virtual memory became readily available on all 370
machines. they were also joined by tymshare offering commercial
vm370-based time-sharing services. lots of post posts mentioning
cp/cms and vm/cms based time-sharing services
http://www.garlic.com/~lynn/submain.html#timeshare

for a little dbms tie-in .... in addition to original relational/sql
system/r being done on vm
http://www.garlic.com/~lynn/submain.html#systemr

there was also the ramis, nomad, focus genre that were done on vm/cms
... misc past posts
http://www.garlic.com/~lynn/2003d.html#15 CA-RAMIS
http://www.garlic.com/~lynn/2003d.html#17 CA-RAMIS
http://www.garlic.com/~lynn/2003n.html#12 Dreaming About Redesigning SQL
http://www.garlic.com/~lynn/2004e.html#15 Pre-relational, post-relational, 1968 CODASYL "Survey of Data  Base Systems"
http://www.garlic.com/~lynn/2004l.html#44 Shipwrecks
http://www.garlic.com/~lynn/2006k.html#35 PDP-1
http://www.garlic.com/~lynn/2006k.html#37 PDP-1
http://www.garlic.com/~lynn/2007e.html#37 Quote from comp.object

and a couple wiki refs:
http://en.wikipedia.org/wiki/Nomad_software
http://en.wikipedia.org/wiki/FOCUS
http://en.wikipedia.org/wiki/Ramis_software

one of the things that started to show up with interactive computing was
games. in the 70s, tymshare had ported the adventure dec fortran source
to vm/cms and i obtained a copy for internal corporate distribution. at
one point, the executives in STL complained that they thot nearly
everybody was spending their days playing adventure on vm/cms (instead
of doing dbms development). recent post mentioning adventure:
http://www.garlic.com/~lynn/2007g.html#0 10 worst PCs

in addition to the local internal corporate datacenters providing
local vm/cms users interactive computing ... there was also the
world-wide HONE system
http://www.garlic.com/~lynn/subtopic.html#hone

providing online, interactive services for sales, marketing and field
people world-wide ... recent reference
http://www.garlic.com/~lynn/2007i.html#77 Sizing CPU

starting out with most of the application written in cms\apl on cp67 and
then transitioning to vm370 and apl\cms. Sometime in the early to mid
70s, sales couldn't even submit a 370 order w/o it having first being
processed by a HONE "configurator".

To continue the theme that there were a lot of cms personal computing
applications ... the type of things you started seeing moving to PCs in
the 80s ... there were a lot of APL-based applications doing "what-if"
type things ... that you later see implemented with PC-based
spreadsheets.

An early one was when the cambridge science center first ported apl\360
to cms for cms\apl. The apl\360 service offerings typically limited
workspace sizes to 16kbytes (or in some cases 32k). With cms\apl, the
size of the APL workspace could be nearly as large as the virtual
address space. This opened up a lot of "what-if" applications using
real-world data. Early on, we found corporate business planners shipping
tapes to Cambridge with the most sensitive customer and corporate
business data ... and using the science center cms\apl facilities to run
"what-if" business planning scenarios. a couple recent references
mentioning business planners in corporate hdqtrs using the cambridge
cms\apl facilities:
http://www.garlic.com/~lynn/2007i.html#20 Does anyone know of a documented case of VM being penetrated by hackers?
http://www.garlic.com/~lynn/2007i.html#77 Sizing CPU

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Fri, 04 May 2007 13:16:20 -0600

Greg Menke <gdmnews@toadmail.com> writes:

There is a very big difference between interrupting a cpu and
arbitrating writing a cache-line back to ram, the latter being what
multiprocessors generally do.  When there is no contention between
cache-lines, then there is no delay.

maybe there is some reference about cross-cache chatter ... which
doesn't show up with processor interrupts visable to software.

the 370 SMPs ... would slow down the processor machine cycle by ten
percent ... to allow for cross-cache chatter ... and any actual
cross-cache chatter (needing handling) would slow the processors down
even further. basically, anytime that cache-line was obtained for
alternation (whether store-thru cache implementation or a store-into
cache implementation) ... there would have to be cross-cache invalidates
to clear cache-line from any other processor.

the cross-cache overhead tripled going from two-processor smp
configuration (one other cache to listen for) to four-processor smp
(three other caches to listen for). in the 3084 time-frame, major
operating systems had some amount of kernel storage re-organization to
force kernel storage to be aligned on cache-line boundaries and
allocated in multiples of cache-line units. this change supposedly
improved overall avg.  system thruput by 5-6percent (just eliminating
possible kernel storage cache-line trashing, aka two different kernel
storage areas overlapping in the same cache-line).

there have been some number of optimizations to eliminate cross-cache
chatter overhead and allow better processor scaling. an early one was
snoopy caches ... where all caches would watch all memory bus operations
... whenever one cache saw a memory bus operation from some other cache
... that involved a cache-line that it had loaded, it would invalidate
it (aka sort of an implicit cross-cache signal).

note this is a cache function ... involving copies of the same data in
different physical caches ... not a processor function. when you have
multiple processors sharing the same physical cache ... the issue
doesn't come up.

lots of past posts mentioning SMP
http://www.garlic.com/~lynn/subtopic.html#smp

a couple recent posts mentioning cross-cache chatter/overhead:
http://www.garlic.com/~lynn/2007d.html#63 Cycles per ASM instruction
http://www.garlic.com/~lynn/2007g.html#16 What's a CPU second?

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 09:19:06 -0600

jmfbahciv writes:

My experience with unions is they order you to work less and less
and less.  My Dad got severely spoken to (this was a Teamster
union) for picking up a broom and sweeping his work area.
Only the broom union members were allowed to touch a broom.

I was brought in during summer break as a full-time employee to help
setup BCS (boeing computer services) operation (earlier they had talked
me into giving a 40hr dataprocessing class during spring break to some
of the BCS technical staff).

This was shortly after BCS had been formed ... it was sort of to move
all dataprocessing into a separate business unit and change it from
expense category and give it profit/loss responsibility (even if it was
mostly internal corporate fund transfers ... however, it was also
allowed to perform dataprocessing and consulting outside the company).

I ran into a similar situation when I suggested that I was going to
install a pencil sharpner on the wall near my desk (this involved
driving screws into the wall).

For other topic drift ... later at the science center
http://www.garlic.com/~lynn/subtopic.html#545tech

there was the son of one of the DNA guys on assignment ... who was heavy
into APL programming at the science center ... after the port of apl\360
to cms\apl had been done ... some recent posts mentioning cms\apl:
http://www.garlic.com/~lynn/2007b.html#32 IBMLink 2000 Finding ESO levels
http://www.garlic.com/~lynn/2007d.html#64 Is computer history taugh now?
http://www.garlic.com/~lynn/2007g.html#31 Wylbur and Paging
http://www.garlic.com/~lynn/2007g.html#48 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007h.html#62 sizeof() was: The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007i.html#20 Does anyone know of a documented case of VM being penetrated by hackers?
http://www.garlic.com/~lynn/2007i.html#77 Sizing CPU
http://www.garlic.com/~lynn/2007j.html#13 Interrupts
http://www.garlic.com/~lynn/2007j.html#17 Newbie question on table design

he left and joined BCS consulting. I dropped by later and he was
explaining how they had a contract with USPS to do the financial
modeling justifying postal stamp increase (from 8 to 10?) ... as I've
often repeated before ... back then, APL was commoningly used for a lot
of things that are currently done with spreadsheets.

for other drift ... a couple recent URLs:

IBM to Layoff Half of Global Services Division
http://slashdot.org/articles/07/05/04/1826221.shtml
Lean and Mean: 150,000 U.S. layoffs for IBM?
http://www.pbs.org/cringely/pulpit/2007/pulpit_20070504_002027.html
IBM will erase 100,000 workers - man with one name
http://www.theregister.com/2007/05/05/ibm_cringely_100k/

Slightly drifting back to the topic ... business trips to various
european plant sites in the early 70s ... recent posts with some mention
http://www.garlic.com/~lynn/2007g.html#47 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#48 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#62 The Perfect Computer - 36 bits?

when I noticed beer in the plant site vending machines ... it was
explained that it was in the union contract (also wine/beer in various
company cafeterias)

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 09:41:47 -0600

krw <krw@att.bizzzz> writes:

If the company goes under because of the retirement and medical plans
the employee isn't helped either.

what i was told in the steel industry case ... was that companies were
declaring bankruptcy when they showed that half the cost of a ton of
steel was retirement benefits.

the problem seems to drift into the theme of the comptroller general
claiming that nobody is able anymore of doing simple middle school
arithmatic ... i.e. the benefits were not fully funded ... but were
being paid out of the current operating revenue. The problem extended
to executives, unions, gov. officials, and employees. Everybody was
taking all the money they could at the moment ... and tomorrow was
somebody else's problem.

Supposedly when the company went into the red in 1992 ... they also
took a charge-off to move to fully funded retirement program
(i.e. since they were already in the red, going further into the red
didn't make that much difference). However, a lot of corporations
never have bothered to do that ... assuming that it in the worst case,
it can be left up to the taxpayers to cover the difference at some
point in the future. An issue ... also raised by the comptroller
general, is that it assumes the number of taxpayers continue to
significantly outnumber the recipients ... there is also a matter of
simple middle school arithmatic involved here too.

past posts mentioning comptroller general and his observation about the
lack of even middle school arithmatic skills:
http://www.garlic.com/~lynn/2006f.html#41 The Pankian Metaphor
http://www.garlic.com/~lynn/2006f.html#44 The Pankian Metaphor
http://www.garlic.com/~lynn/2006o.html#61 Health Care

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 09:56:02 -0600

somewhat tieing together these two posts
http://www.garlic.com/~lynn/2007j.html#19 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#20 IBM Unionization

maybe there is requirement to return to using APL for doing financial
modeling ... of course it still doesn't really make up for the lack of
middle school arithmatic skills ... since it would take some
understanding to even write the APL programs.

lots of past posts mentioning APL (and/or HONE):
http://www.garlic.com/~lynn/subtopic.html#hone

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 10:36:16 -0600

re:
http://www.garlic.com/~lynn/2007j.html#19 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#20 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#21 IBM Unionization

other folklore ... i've seen lots of references ... but haven't
actually read the statutes ... claims that in the mid-90s congress
passed some legislation that changed some accounting rules, allowing
companies to claim funded retirement plans as assets. it doesn't make
a whole lot of practical difference to the running of the company
... except for possibly the size of executive bonuses that are tied to
such numbers. however, there is the claim that since it is claimed as
an asset, if a company were to declare bankruptcy ... assets are
available for paying off creditors (possibly even including funded
retirement benefits).

for other drift ... in the mid-80s (20+ yrs ago), I spent a summer
touring around europe doing market support, giving talks and teaching
classes (I wasn't paying close attention to geography and one sequence
went boeblingen, stockholm, zurich over 3week period when it would
have been much easier to do boeblingen, zurich, stockholm). At one
location where i was teaching a 40hr, one week class, i was in the
habit of coming in an hr or two early and staying a couple hrs in the
evening to login back to the states and do email and some number of
other things.

i then asked if i could come in over the weekend to get some work
done. i was told that i was creating a significant bureaucratic paper
work problem for the local management. It seems that the company
benefits had included vacation time that was one week more than the
gov. mandated.  As a result, the employees had come to think of
themselves as priviledge compared to their fellow countrymen.

Recently, the gov. had just increased the mandated vaction time by one
week and many employees believed that the company should then also
increase its vacation time by a week (to preserve the one week more
than gov. mandated for all other workers in the country). Since the
company hadn't done that, the unions were taking various work
actions. This included the guard union observing that I was spending
more than 8hrs/day in the bldg ... and filing various forms with the
gov. about work violations (which management had to respond to
... also in writting with the appropriate forms). If I were to (also)
come in on the weekend, it would significantly increase the
gov. paperwork forms that they would have to file (in response to the
anticipated forms that would be filed by the guard union).

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 12:42:16 -0600

Greg Menke <gdmnews@toadmail.com> writes:

Microsoft software runs on single or multiple cpu systems.  Win9x and
earlier won't take advantage of multiple processors, IIRC extra
processors are idle.  You seem preoccupied with syncronization methods
but few minutes using google will show you that this is a well-solved
problem.  Yes, its true that if there are no measures taken to serialize
access to shared resources that the processors will step on each other's
work- but its also true that serializing them is easily done.  Google
compare and swap insructions.

or check some principle of operations ... from 370 onwards. Charlie
invented compare&swap instruction when he was working on fine-grain
multiprocessor locks for cp67 at the science center
http://www.garlic.com/~lynn/subtopic.html#545tech

the mnemonic CAS comes from charlie's initials ... which then required
coming up with the phrase compare and swap, to go with
charlie's initials.

as previously mentioned when attempting to first get compare&swap into
370 architecture ... it was rejected based on various assertions that
test&set instruction was sufficient for handling multiprocessor
syncronizations. the challenge given was that in order to justify
compare&swap instruction inclusion in 370 architecture
... there had to be uses that weren't multiprocessor specific; thus
were born the descriptions of using compare&swap
instruction in multithreaded/multiprogramming applications ... whether
or not running in a multiprocessor environment. a few recent post
references:
http://www.garlic.com/~lynn/2007i.html#31 Latest Principles of Operation
http://www.garlic.com/~lynn/2007i.html#49 John W. Backus, 82, Fortran developer, dies

above post in this thread also includes URL pointer to the description
of compare&swap uses in the current principles of operation
(some 35yrs later)

lots of past posts mentioning multiprocessor and/or compare&swap
instruction
http://www.garlic.com/~lynn/subtopic.html#smp

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory
Date: Sat, 05 May 2007 13:03:33 -0600

paul c <toledobythesea@oohay.ac> writes:

I had the attitude that "batch" was a synonym for "bait and
switch". Marketing material would show people getting instant answers
to various questions but once the new machinery was installed,
customers would complain that there weren't hours in the day to get
all their work done. Service rep's would then advise that productivity
would improve with better utilization, eg., keeping the expensive cpu
100% busy, eliminate interactive work, make people schedule their
machine time in advance and jostle for position in the queue.  Soon a
scheduling software industry appeared.  Accountants saw the
opportunity to record and ration machine time.  Even when the TRS80's,
Apples and PC's appeared, nostalgia for those ways didn't evaporate,
maybe even grew.  Human nature to preserve the old ways and
idiosyncrasies of most fields.  It will probably take decades for SQL
to go away.

re:
http://www.garlic.com/~lynn/2007j.html#16 Newbie question on table design.

note as somewhat implied in this previous post
http://www.garlic.com/~lynn/2007j.html#14 Newbie question on table design.

the development tools used by SJR for system/r and sql
http://www.garlic.com/~lynn/2007d.html#17 Jim Gray is Missing

weren't nearly as sophisticated or powerful as the tools used by the
tools group in the los gatos VLSI lab for their DBMS development.
Somewhat as a result, the tools group also had somewhat more
experience and sophistication in doing a DBMS language for their
implementation.

for topic other drift ... a reference about early on, somebody doing some
cms application re-implementation for the TRS80
http://www.garlic.com/~lynn/2007.html#1 "The Elements of Programming Style"

for slightly other drift ... GML
http://www.garlic.com/~lynn/submain.html#sgml

had been invented at the science center
http://www.garlic.com/~lynn/subtopic.html#545tech

in 1969 (aka G, M, and L, are the initials of the three people
responsible ... they then had to come up with the phrase "generalized
mark-up language" to correspond with their initials). It then took
quite awhile for GML to evolve thru SGML, HTML, XML, etc.

GML language support had been added to the CMS document formating
application "SCRIPT". Archeological reference to HTML evolving from
the "waterloo" version of CMS script command in use at CERN:
http://infomesh.net/html/history/early/

other historical trivia, the first webserver outside europe was on
vm/cms system at slac:
http://www.slac.stanford.edu/history/earlyweb/history.shtml

for other trivia ... the current w3c offices are just around the
corner from the old science center location at 545 tech. sq.

misc. past posts mentioning the "waterloo" version of the CMS script command:
http://www.garlic.com/~lynn/2002b.html#46 ... the need for a Museum of Computer Software
http://www.garlic.com/~lynn/2004l.html#72 Specifying all biz rules in relational data
http://www.garlic.com/~lynn/2004l.html#74 Specifying all biz rules in relational data
http://www.garlic.com/~lynn/2005e.html#34 Thou shalt have no other gods before the ANSI C standard
http://www.garlic.com/~lynn/2005f.html#49 Moving assembler programs above the line
http://www.garlic.com/~lynn/2005l.html#10 who invented CONFIG/SYS?
http://www.garlic.com/~lynn/2005q.html#3 winscape?
http://www.garlic.com/~lynn/2005s.html#12 Flat Query
http://www.garlic.com/~lynn/2006d.html#35 Fw: Tax chooses dead language - Austalia
http://www.garlic.com/~lynn/2006m.html#55 The System/360 Model 20 Wasn't As Bad As All That
http://www.garlic.com/~lynn/2006p.html#16 Comic Sans was Re: An alternative history
http://www.garlic.com/~lynn/2007d.html#29 old tapes

a little dbms folklore

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: a little dbms folklore
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 15:34:00 -0600

from recent thread over in comp.databases.theory (with standard amount
of topic drift):
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#8 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#9 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#11 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#12 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#14 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#16 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#17 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#24 Newbie question on table design

Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 16:40:19 -0600

ArarghMail705NOSPAM writes:

No.  Core dumps don't mean much unless you already have a pretty good
clue as to what is where in core.

past posts mentioning dump analyzer program that i wrote in REX(X)
http://www.garlic.com/~lynn/submain.html#dumprx

objective was in 3mths elapsed time working half-time ... have something
with ten times the function and ten times the performance of the
existing tool (that had been implemented in assembler). it included some
amount of code that searched the dump for specific kinds of failure
signatures.

recent posts with references
http://www.garlic.com/~lynn/2007i.html#30 John W. Backus, 82, Fortran developer, dies
http://www.garlic.com/~lynn/2007i.html#43 Latest Principles of Operation

in cp67, "svc 0" instruction (supervisor call with function code zero)
was used to invoke the kernel failure & automatic dump process. they
were sprinkled around the kernel to invoked falure/dump when various
conditions were encountered. one of the enhancement for cp67 "release 3"
was to add automatic reboot/reipl after the dump had been taken. recent
posts mentioning cp67 automatic reboot
http://www.garlic.com/~lynn/2007d.html#22 How many 36-bit Unix ports in the old days?
http://www.garlic.com/~lynn/2007d.html#23 How many 36-bit Unix ports in the old days?
http://www.garlic.com/~lynn/2007g.html#44 1960s: IBM mgmt mistrust of SLT for ICs?
http://www.garlic.com/~lynn/2007i.html#21 John W. Backus, 82, Fortran developer, dies

in the morph from cp67 to vm370, i suggested that each of the svc0
instructions be followed by unique identification codes to facilitate
identifying the failure cause. The 1st response to the suggestion was
that you can't embed data in an instruction stream ... since there would
be an execution failure on return to the svc0 instrucation. Pause ...
there was never going to be a return to the svc0 instruction ... since
the kernel was going to dump and fail.

Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 17:29:28 -0600

Frank McCoy <mccoyf@millcomm.com> writes:

Ten times the function and performance?
Usually I've been quite satisfied to get three to four times the
performance; while management usually expected only 1.5 to 2 times ...
Thus making me look pretty good when the program got finished.

re:
http://www.garlic.com/~lynn/2007j.html#26 Even worse than UNIX

it was somewhat a given that it might be possible to do ten times the
function ... with appropriate higher-level language ... as compared to
assembler code implementation ... the trick was also getting ten times
the performance out of the interpreted REXX implementation compared to
the assembler implementation ... part of the demonstration was to
support a claim that freed from having to deal with low-level
assembler gorp, it allowed a person to better concentrate on solving
the problem ... which also included being able to recognize new ways
of optimized implementation.

re:
http://www.garlic.com/~lynn/submain.html#dumprx

early on in working on cp67 as undergraduate ... i had been known to
achieve 100 times performance improvement for some kernel pathlengths.
but over the yrs ... as more and more things were optimized ... it
became harder and harder to find additional things where it was
possible to obtain 100 times performance improvement.

Even worse than UNIX

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Even worse than UNIX
Newsgroups: alt.folklore.computers
Date: Sat, 05 May 2007 17:42:51 -0600

Frank McCoy <mccoyf@millcomm.com> writes:

Ten times the function and performance?
Usually I've been quite satisfied to get three to four times the
performance; while management usually expected only 1.5 to 2 times ...
Thus making me look pretty good when the program got finished.

re:
http://www.garlic.com/~lynn/2007j.html#26 Even worse than UNIX
http://www.garlic.com/~lynn/2007j.html#27 Even worse than UNIX

for totally different subject ... about a decade ago, we had been
asked to look at ROUTES ... one of the major applications in airline
res system ... and given a list of ten things that were then currently
impossible to do.

looking at it, completely changed the paradigm of how ROUTES went
about being implemented ... and were also able to then implement all
ten impossible things. As part of the complete paradigm change
... also speeded up things by a factor of 100 times.

However, also as part of the paradigm change ... three separate
person/human transactions were collapsed into a single operation
... and that single operation did quite a bit additional work
(compared to what had happened in all three of the previous separate
operations) ...  so it actually netted out to have only ten times the
"transaction" rate as the prior implementation (although the
definition of what was now a transaction had dramatically changed).

a couple recent posts mentioning having redone ROUTES:
http://www.garlic.com/~lynn/2007g.html#22 Bidirectional Binary Self-Joins
http://www.garlic.com/~lynn/2007g.html#41 US Airways badmouths legacy system
http://www.garlic.com/~lynn/2007h.html#41 Fast and Safe C Strings: User friendly C macros to Declare and use C Strings

note initially ... it was only about 20 times faster ... but careful
study of the cache structure and re-arranging a lot of program data
structures to better match the machine's cache operation ... got another
five times improvement (5*20 ... netted 100 times).

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 08:07:38 -0600

Andrew Swallow <am.swallow@btopenworld.com> writes:

I first programmed a double CPU minicomputer in 1979.  Ten years later
microprocessors had allowed the electronic engineers to get 6 CPUs on
the same shelf.  Each CPU had its own software.  The software to
control the radios was definitely different from the software to run
the MMI. However they all had to work together to get what we now call
email out.

most microprogrammed 360s & 370s had physical separate boxes for
different microprocessors ... when there were different
microprocessors ... i.e.  processors, channels, controllers, devices
... example is 370/158 with microprocessor ... but two different
microcode loads ... one for 370 execution and one for integrated
channel operation. i've mentioned before that transition to 303x
... involved repackaging a 158 microprocessor engine (this is not so
much micro as in small but micro as in microcode) as (dedicated)
"channel director" (with only the integrated channel microcode
load). Then a 3031 become two 370/158 microprocessors ... one with
just the 370 microcode load and the "channel director" with just the
integrated channel microcode load.

however, in the early 70s, Boeblingen got into a little trouble (with
corporate) doing something different for the 370/115 and 370/125. They
created a 9-position shared memory bus (i.e. positions for up to nine
processors). A 370/115 was an microengine running 370 microcode load
able to execute 370 at about 80kips ... and 2-8 other (identical
microengines) with various microcode loads to perform various kinds of
i/o controller functions. A 370/125 was identical to a 370/115
... except the microengine executing 370 microcode load was about 50%
faster, able to execute 370 instructions at about 120kips.

The microengines executed an avg. 10 native instructions for every 370
instruction ... so the native performance of the 115 engine was around
800kips and the native performance of the 125 engine was around
1.2mips.

since there was no processor caches (this were slower microprocessors
that didn't have the performance mismatch between processor speed and
memory speed) ... and the system didn't run similar microcode loads
... there was less problem with multiprocessor coordination
... although the processors executing i/o controller functions were
responsible for "executing" channel programs ... which could also be
operated on by the processor running 370 microcode.

recent posts mentioning Boeblingen lab
http://www.garlic.com/~lynn/2007c.html#14 How many 36-bit Unix ports in the old days?
http://www.garlic.com/~lynn/2007g.html#47 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#48 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#62 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007j.html#22 IBM Unionization

recent posts mentioning executing (360/370) "channel programs"
http://www.garlic.com/~lynn/2007d.html#28 SVCs
http://www.garlic.com/~lynn/2007e.html#14 Cycles per ASM instruction
http://www.garlic.com/~lynn/2007e.html#17 A way to speed up level 1 caches
http://www.garlic.com/~lynn/2007e.html#19 Cycles per ASM instruction
http://www.garlic.com/~lynn/2007e.html#27 IBM S/360 series operating systems history
http://www.garlic.com/~lynn/2007e.html#38 FBA rant
http://www.garlic.com/~lynn/2007e.html#41 IBM S/360 series operating systems history
http://www.garlic.com/~lynn/2007e.html#46 FBA rant
http://www.garlic.com/~lynn/2007f.html#0 FBA rant
http://www.garlic.com/~lynn/2007f.html#4 ISAM and/or self-modifying channel programs
http://www.garlic.com/~lynn/2007f.html#6 IBM S/360 series operating systems history
http://www.garlic.com/~lynn/2007f.html#20 Historical curiosity question
http://www.garlic.com/~lynn/2007f.html#33 Historical curiosity question
http://www.garlic.com/~lynn/2007f.html#34 Historical curiosity question
http://www.garlic.com/~lynn/2007g.html#23 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007h.html#3 21st Century ISA goals?
http://www.garlic.com/~lynn/2007h.html#7 The Mainframe in 10 Years
http://www.garlic.com/~lynn/2007h.html#9 21st Century ISA goals?
http://www.garlic.com/~lynn/2007i.html#20 Does anyone know of a documented case of VM being penetrated by hackers?

recent posts mentioning 303x channel director
http://www.garlic.com/~lynn/2007b.html#18 How many 36-bit Unix ports in the old days?
http://www.garlic.com/~lynn/2007d.html#21 How many 36-bit Unix ports in the old days?
http://www.garlic.com/~lynn/2007d.html#62 Cycles per ASM instruction
http://www.garlic.com/~lynn/2007e.html#32 I/O in Emulated Mainframes
http://www.garlic.com/~lynn/2007f.html#28 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007f.html#65 History - Early Green Card
http://www.garlic.com/~lynn/2007g.html#17 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#23 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#57 IBM to the PCM market(the sky is falling!!!the sky is falling!!)
http://www.garlic.com/~lynn/2007h.html#1 21st Century ISA goals?
http://www.garlic.com/~lynn/2007i.html#31 Latest Principles of Operation

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 08:33:35 -0600

Andrew Swallow <am.swallow@btopenworld.com> writes:

BAH we are not talking about a 5 ton Cray from 30 years ago but a
single chip smaller than a finger nail.  One microprocessor chip
containing 80 X86 family CPUs (cores).  It is designed to go into
standard home PCs in 2 years time.

... er, almost

Intel pledges 80 cores in five years
http://news.com.com/Intel+pledges+80+cores+in+five+years/2100-1006_3-6119618.html

from above:

Intel's prototype uses 80 floating-point cores, each running at 3.16GHz,
Justin Rattner, Intel's chief technology officer, said in a speech
following Otellini's address. In order to move data in between
individual cores and into memory, the company plans to use an on-chip
interconnect fabric and stacked SRAM (static RAM) chips attached
directly to the bottom of the chip, he said.

... snip ...

the prototype wasn't full x86 processors

The Era of Tera: Intel Reveals more about 80-core CPU
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2925

now Rapport has 256 core and moving to "kilocore" ... i.e. 1024 cores.

IBM, Rapport's Kilocore, and reconfigurable computing
http://arstechnica.com/news.ars/post/20060407-6556.html

and with a little help from IBM becomes 1025 cores ... with the addition
of a single powerpc core.

but as per this post ... there is significant barrier with regard to
existing programming paradigms and being able to effectively utilize the
cores:
http://www.garlic.com/~lynn/2007i.html#78 John W. Backus, 82, Fortran developer, dies

also had ref to last yrs hotchips ... which had a number of
presentations by cellphone companies about high-performance, low-power
multicore chips ... able to process highly compressed full-motion
video.

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 09:35:58 -0600

jmfbahciv writes:

Starting rate was $29/hour....but I can't recall the year.

which may be the salary ... not the fully loaded rate. i seem to
remember something from '90s about usps entry fully loaded rate
(salary, taxes, benefits, etc) around $30/hr ... slight topic drift,
post mentioning (old) usps financial modeling
http://www.garlic.com/~lynn/2007j.html#19 IBM Unionization

this gao report
http://www.gao.gov/new.items/d06285.pdf

has a lot of stuff sliced and diced in a number of different ways for
salary/wages, benefits, fully loaded compensation for period 1991 to
2005. For instance, avg total (fully loaded) compensation in 2005 for
all workers in companies with >500 workers was $33.48/hr.

so foreign automobile companies did a couple different things when
import quotas were instituted. one was move from cars at the low-end
to cars at the high-end (i.e. the import quotas were on the number of
cars ... so change from selling cars less than domestic vehicles to
selling same number of cars at the high-end).

the other tactic was start building cars in the US. one of the
comments from the time was that they were use to being able to hire
high-school graduates to perform the work. however, in the US they
were finding they needed to require at least 2yrs college (junior
college degree) in order to get people with at least high-school
education.

other posts in this thread:
http://www.garlic.com/~lynn/2007j.html#20 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#21 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#22 IBM Unionization

misc. past posts mentioning foreign automobiles ... and/or census
studies finding that at least half the high-school graduate age people
were functionally illiterate
http://www.garlic.com/~lynn/2000f.html#43 Reason Japanese cars are assembled in the US (was Re: American bigotry)
http://www.garlic.com/~lynn/2002f.html#43 Foreign Cars (was: Computers in Science Fiction)
http://www.garlic.com/~lynn/2002k.html#45 How will current AI/robot stories play when AIs are real?
http://www.garlic.com/~lynn/2003i.html#28 Offshore IT
http://www.garlic.com/~lynn/2003i.html#45 Offshore IT
http://www.garlic.com/~lynn/2003i.html#55 Offshore IT
http://www.garlic.com/~lynn/2003i.html#61 TGV in the USA?
http://www.garlic.com/~lynn/2003p.html#33 [IBM-MAIN] NY Times editorial on white collar jobs going
http://www.garlic.com/~lynn/2004b.html#42 The SOB that helped IT jobs move to India is dead!
http://www.garlic.com/~lynn/2004c.html#51 [OT] Lockheed puts F-16 manuals online
http://www.garlic.com/~lynn/2004d.html#18 The SOB that helped IT jobs move to India is dead!
http://www.garlic.com/~lynn/2004h.html#18 Low Bar for High School Students Threatens Tech Sector
http://www.garlic.com/~lynn/2004h.html#22 Vintage computers are better than modern crap !
http://www.garlic.com/~lynn/2005e.html#48 Mozilla v Firefox
http://www.garlic.com/~lynn/2005g.html#43 Academic priorities
http://www.garlic.com/~lynn/2005s.html#2 Internet today -- what's left for hobbiests
http://www.garlic.com/~lynn/2006.html#23 auto industry
http://www.garlic.com/~lynn/2006.html#44 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006g.html#14 The Pankian Metaphor
http://www.garlic.com/~lynn/2006g.html#20 The Pankian Metaphor
http://www.garlic.com/~lynn/2006l.html#63 DEC's Hudson fab
http://www.garlic.com/~lynn/2006m.html#49 The Pankian Metaphor (redux)
http://www.garlic.com/~lynn/2006v.html#14 In Search of Stupidity
http://www.garlic.com/~lynn/2007f.html#50 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#7 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007g.html#29 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#34 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007g.html#52 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007i.html#13 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007i.html#24 John W. Backus, 82, Fortran developer, dies
http://www.garlic.com/~lynn/2007i.html#79 John W. Backus, 82, Fortran developer, dies

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 10:33:35 -0600

jmfbahciv writes:

Then it's not really an SMP system if there is only one bus.
That makes the bus the single-point error.

SMP traditional stands for either 1) Symmetrical MultiProcessing or 2)
Shared-Memory Processing (implying multiple processors sharing the
same memory).

some vendors would market SMPs (only) with redundant and/or
partionable configurations. for instance, in the 60s, for the 360
two-processor SMPs, it was always possible to split them and run as
two, independent single processor machines. this continued thru the
370s up until the 3081 ... where they attempted to introduce to the
term "dyadic" to differentiate it from the earlier configurations that
could be partitioned into independent operations.

whether or not a particular vendors SMP was error-tolerate or not was
more by convention than part of the industry definition of SMP.

recent post mentioning DEC press release moving from asymmetrical
multiprocessing to symmetrical multiprocessing support (circa spring
'88) ... i.e. they may have been SMP in the "shared-memory processing"
sense ... but hadn't been SMP in the "symmetrical multiprocessing"
sense:
http://www.garlic.com/~lynn/2007.html#46 How many 36-bit Unix ports in the old days?

lots of old posts mentioning smp and/or compare&swap instruction
http://www.garlic.com/~lynn/subtopic.html#smp

not that this was also referred to as "tightly-coupled" in mainframe
circles ... to differentiate from "loosely-coupled" (i/o sharing but
not memory sharing). tighly-coupled tended to be less fault resistant
because common code (in common memory) could result in failures ...
recent post mentioning study that indicated starting at least in the
early 80s, hardware represented smaller and smaller portion of the
cause of failures:
http://www.garlic.com/~lynn/2007h.html#76 John W. Backus, 82, Fortran developer, dies

as previously mentioned ... at one point, my wife had been conned into
going to pok to be in charge of loosely-couple architecture ... where
she originated peer-coupled shared data architecture
http://www.garlic.com/~lynn/submain.html#shareddata

later we did the ha/cmp product
http://www.garlic.com/~lynn/subtopic.html#hacmp

and coined the terms disaster survivability and geographic
survivability (to differentiate from disaster recovery).
http://www.garlic.com/~lynn/submain.html#available

i.e. loosely-coupled and/or cluster implementations are somewhat
easier to create geographically distributed failure recovery
implementations. Also, as both hardware and software have gotten more
reliable ...  environmental problems/failures rise to the top of the
list of common source of outages ... and as cost of hardware and
transmission bandwidth have fallen, more and more services find that
they can justify geographic distributed failure recovery (for
continuous availability).

other topic drift ... old email referencing working on ha scaleup
http://www.garlic.com/~lynn/lhwemail.html#medusa

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 11:21:40 -0600

jmfbahciv writes:

Starting rate was $29/hour....but I can't recall the year.

re:
http://www.garlic.com/~lynn/2007j.html#31 IBM Unionization

oh ... and these posts
http://www.garlic.com/~lynn/2000f.html#41 Reason Japanese cars are assembled in the US (was Re: American bigotry)
http://www.garlic.com/~lynn/2001d.html#43 Economic Factors on Automation
http://www.garlic.com/~lynn/2004b.html#52 The SOB that helped IT jobs move to India is dead!
http://www.garlic.com/~lynn/2004h.html#22 Vintage computers are better than modern crap !
http://www.garlic.com/~lynn/2005s.html#2 Internet today -- what's left for hobbiests
http://www.garlic.com/~lynn/2006.html#23 auto industry
http://www.garlic.com/~lynn/2006g.html#14 The Pankian Metaphor
http://www.garlic.com/~lynn/2006g.html#17 The Pankian Metaphor
http://www.garlic.com/~lynn/2006g.html#20 The Pankian Metaphor
http://www.garlic.com/~lynn/2006m.html#49 The Pankian Metaphor (redux)

mentions an old article calling for 100percent tax on (massive)
unearned profits in the automobile industry in the wake of the import
quotas.

the article claimed that the purpose of the import quotas was to give
the domestic automobile industry breathing room to remake themselves
into a competitive body. the reduction in cheap imports would lesson
the downward pressure on prices charged for domestic cars ... giving
them increased profits which could be funneled into remaking their
business.  The article observed that instead, the (significant)
increased profits went into increases in workers and executives
salaries/bonuses and stock dividends.

There were also secondary effects ... with the foreign importers
observing that they could sell as many high-end cars as low-end cars
... totally changed the type of car they sold. This almost totally
eliminated the downward pressure on domestic car prices ... supposedly
allowing domestic manufacturers to nearly double the price of the same
automobile over a period of a few yrs.

There were other secondary effects ... the motivation to the foreign
importers to totally change the type of car they sold ... also
contributed to changing how they did car development ... cutting the
traditional 7-8yrs elapsed time to 2-3yrs. Some of the effects of this
could be eventually seen in the early 90s behind the "C4" effort in
the domestic car industry, which was targeted at leveraging
(primarily) IT/dataprocessing technology to cut the traditional 7-8yrs
lead time to come out with a new model (aka yr to yr changes tend to
be mostly cosmetic) to 3yrs (in order to finally(?)  compete with
foreign operations).

one of the issues behind C4 ... is that as long as car buying habits
remain relatively static ... there isn't much to differentiate 7-8yr
development cycle and 2-3yr development cycle. However, when buying
habits go into same amount of flux and change ... being able to
respond to what the customers are buying in 2-3yrs (or less) gives a
significant advantage over organizations that take 7-8yrs to respond.

past posts mentioning C4 effort:
http://www.garlic.com/~lynn/2000f.html#43 Reason Japanese cars are assembled in the US (was Re: American bigotry)
http://www.garlic.com/~lynn/2003i.html#61 TGV in the USA?
http://www.garlic.com/~lynn/2004c.html#51 [OT] Lockheed puts F-16 manuals online
http://www.garlic.com/~lynn/2004h.html#22 Vintage computers are better than modern crap !
http://www.garlic.com/~lynn/2006.html#44 IBM 610 workstation computer
http://www.garlic.com/~lynn/2006g.html#14 The Pankian Metaphor
http://www.garlic.com/~lynn/2006g.html#20 The Pankian Metaphor
http://www.garlic.com/~lynn/2006m.html#49 The Pankian Metaphor (redux)
http://www.garlic.com/~lynn/2006v.html#14 In Search of Stupidity
http://www.garlic.com/~lynn/2007f.html#50 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#29 The Perfect Computer - 36 bits?
http://www.garlic.com/~lynn/2007g.html#34 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007g.html#52 U.S. Cedes Top Spot in Global IT Competitiveness
http://www.garlic.com/~lynn/2007i.html#13 U.S. Cedes Top Spot in Global IT Competitiveness

IBM Unionization

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: IBM Unionization
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 11:30:57 -0600

re:
http://www.garlic.com/~lynn/2007j.html#31 IBM Unionization
http://www.garlic.com/~lynn/2007j.html#33 IBM Unionization

oops, do i hear the name john boyd and OODA-loops being whispered?
http://www.garlic.com/~lynn/subboyd.html

John W. Backus, 82, Fortran developer, dies

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: John W. Backus, 82, Fortran developer, dies
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 11:41:53 -0600

John Ahlstrom <AhlstromJK@comcast.net> writes:

What kind of trouble with corporate and why?

re:
http://www.garlic.com/~lynn/2007j.html#29 John W. Backus, 82, Fortran developer, dies

they weren't suppose to do multiple microprocessor common bus
implementation. it is now long ago and far away (35yrs) ... but i have
vague recollection that it may have involved some domestic plants
complaining that Boeblingen had done a more advanced/competitive
design

Display Technologies Evolution

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Display Technologies Evolution
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 13:46:57 -0600

Brian Inglis <Brian.Inglis@SystematicSW.Invalid> writes:

Monochrome plasma panels were also used at one point in products such as
the IBM 3190 Information distributor, which had four times the
resolution of a normal 80x25 screen, and could be used to display 1-4
logical screens, with a maximum single screen capacity up to 132x60 (a
normal lineprinter page at that time).

3290 plasma display predated 3190 ... some search engine use found a
picture/reference:
http://www.recycledgoods.com/ProductDetails.aspx?productID=7774&SubCategoryID=101&CategoryID=1

and

The History of Plasma Displays
http://www.plasmatvscience.org/plasmatv-history3.html

from above:

IBM took an early interest as well, and the lure of Big Blue's prestige
and deep pockets forced Merriam and Alpert into some ticklish
negotiations between the two corporate players, with the happy result
that U of I collected a million dollars from IBM in exchange for another
license. That license would lead in 1983 to the IBM 3290 Information
Panel, "the industry's first mass-produced, large-screen plasma display
terminal for commercial use," according to an IBM advertisement.

... snip ...

strange FTP problem--transatlantic asymmetry

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: strange FTP problem--transatlantic asymmetry
Newsgroups: comp.protocols.tcp-ip
Date: Sun, 06 May 2007 14:42:31 -0600

patrick.peters writes:

What kind of a problem gives you good latency numbers but poor
application performance?  What kind of problem only shows up when the
big data flows move east?

routing doesn't have to be symmetrical, bandwidth/traffic shaping
doesn't have to be symmetrical, individual physical links may not be
symmetrical, contention may not be symmetrical (aka link may be
symmetrical but dominant traffic flow may not be symmetrical, aka lots
of traffic from servers in the west to clients in the east).

traceroute from both ends may give whether the hops are the same
(symmetrical routing)

large packet traceroute (from both ends) may give whether the
transmission latency over each individual hop is symmetrical.

Small packets may give similar latency for similar hop ... even if
bandwidth/traffic of the links are significantly different (or different
in different directions). traceroute with different port numbers might
turn up port-number specific traffic shaping.

bandwidth/traffic shaping can occur at ip layer and may even be
udp/tcp/port specific ... and can be asymmetrical. asymmetrical
bandwidth can also occur at lower levels.

wiki page on traffic shaping
http://en.wikipedia.org/wiki/Traffic_shaping

other traffic shaping reference from search engine
http://www.cisco.com/univercd/cc/td/doc/product/software/ios120/12cgcr/qos_c/qcpart4/qcfrts.htm

any of the possible characteristics might be asymmetrical for any
specific hop.

Problem with TCP connection close

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Problem with TCP connection close
Newsgroups: comp.protocols.tcp-ip
Date: Sun, 06 May 2007 15:09:41 -0600

David Schwartz <davids@webmaster.com> writes:

Unfortunately, TCP takes resources and times to start up and shut down
a connection. There's no a whole lot you can do about this. There are
known serious hazards with trying to reduce the TIME_WAIT interval.

aka TCP session has a minimum seven packet exchange ... plus FINWAIT
tail-end on close.

mid-90s HTTP webservers encountered significant TCP scale-up problem
with common implementations that had linear scan of FINWAIT list. the
problem was that some loaded webservers were cycling TCP sessions so
fast (resulting in quite lengthy FINWAIT lists) that several of the
major webservers were hitting 100% processor busy and spending nearly
all of that scanning the FINWAIT list. Several vendors relatively
quickly put together new releases that significantly reworked how
FINWAIT list was handled.

there has also been work on reliable transactions protocols that had
fewer minimum packet exchange ... i.e. like VMTP/RFC1045 with 5 minimum
packet exchange.

Newbie question on table design

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Newbie question on table design.
Newsgroups: comp.databases.theory,alt.folklore.computers
Date: Sun, 06 May 2007 15:59:14 -0600

Anne & Lynn Wheeler <lynn@garlic.com> writes:

one of the things that started to show up with interactive computing
was games. in the 70s, tymshare had ported the adventure dec fortran
source to vm/cms and i obtained a copy for internal corporate
distribution. at one point, the executives in STL complained that they
thot nearly everybody was spending their days playing adventure on
vm/cms (instead of doing dbms development). recent post mentioning
adventure:
http://www.garlic.com/~lynn/2007g.html#0 10 worst PCs

re:
http://www.garlic.com/~lynn/2007j.html#17 Newbie question on table design

STL (south san jose, dbms and language development, subseqently renamed
the silicon valley lab) and Hursley (UK, communication and cics
development) looked at off-shift dataprocessing offloading in 1980. The
dominant development platform at both locations was vm/cms interactive
computing. like in the reference to palo alto vm/cms this sort of
interactive workload tends to be running at full capacity during first
shift with much lower off-shift use.
http://www.garlic.com/~lynn/2007j.html#13 Interrupts

the strategy was to get a "high-bandwidth" double-hop satellite link
between STL and Hursley (i.e. from west coast up to the geo-sync
satellite over the US, down to the east coast, up again to geo-sync
satellite over the atlantic and down to Hursley). Since there was 8hr
time-difference ... some of the 1st shift Hursley workload could be run
on the STL machines (being 3rd shift in california) and some of the 1st
shift STL workload could be run on Hursley machines (being 2nd shift in
UK).

This was going to be showcase operation ... so some of the strategy
people stepped in to "force" the link to operate with MVS JES2/SNA (even
tho the dominate operation was vm/cms which didn't use SNA for network
links).

They tried to bring up JES2/SNA on the link ... and nothing happened.
Somebody then suggested to try it with VM link just to check things out
... and it came up with no errors. They immediately switched back to
JES2/SNA on the link and nothing happened. The "official" conclusion was
that there was significant transmission errors on the double-hop
satellite link and VM link error handling was too primitive to recognize
the problems.  The actual situation was that the JES2/SNA had built in
round-trip timeout ... which the round-trip double-hop satellite
propagation delay was exceeding (and which they couldn't "fix") ... but
the VM link support did adjust for the round-trip (double-hop satellite)
propagation delay.

for other MVS JES2 topic drift between San Jose and Hursley ... old
post mentioning (unless converted) network traffic between different
versions of MVS JES2 (with incompatible network traffic headers) would
result in JES2 failure that would also bring down the MVS system
http://www.garlic.com/~lynn/2006x.html#8 vmshare

misc. other posts in this thread:
http://www.garlic.com/~lynn/2007j.html#2 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#7 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#8 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#9 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#11 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#12 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#14 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#16 Newbie question on table design
http://www.garlic.com/~lynn/2007j.html#24 Newbie question on table design

past posts mentioning the attempt at San Jose/Hursley off-shift offloading:
http://www.garlic.com/~lynn/99.html#212 GEOPLEX
http://www.garlic.com/~lynn/2002q.html#35 HASP:
http://www.garlic.com/~lynn/2003j.html#19 tcp time out for idle sessions
http://www.garlic.com/~lynn/2004k.html#19 FAST TCP makes dialup faster than broadband?
http://www.garlic.com/~lynn/2004o.html#60 JES2 NJE setup
http://www.garlic.com/~lynn/2004o.html#61 360 longevity, was RISCs too close to hardware?
http://www.garlic.com/~lynn/2006m.html#16 Why I use a Mac, anno 2006
http://www.garlic.com/~lynn/2006s.html#17 bandwidth of a swallow (was: Real core)

Disc Drives

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: Re: Disc Drives
Newsgroups: alt.folklore.computers
Date: Sun, 06 May 2007 16:18:02 -0600

scott@slp53.sl.home (Scott Lurndal) writes:

The real question is what is the MTBF for drives that spin-up/spin-down
vs. the MTBF for drives that stay up all the time, and is the difference
significant.

the previous reference to the google disk MTBF study on 100,000 drives
(there is also mention of a cmu disk MTBF study on 100,000 drives)
http://www.garlic.com/~lynn/2007j.html#10 Disc Drives
and
http://www.garlic.com/~lynn/2007h.html#13 Question on DASD Hardware

doesn't mention spin-up/spin-down frequency related to MTBF ... but does
look at lightly loaded versus heavily loaded. slightly surprising is
that after 3yrs, heavily loaded disks tend to have slightly better MTBF
than lightly loaded disks (with possible explanation that less reliable
heavily loaded disks tended to not make it to 3yrs).

z/VM usability

From: Anne & Lynn Wheeler <lynn@garlic.com>
Subject: re: z/VM usability
Newsgroups: bit.listserv.vmesa-l,alt.folklore.computers
Date: Sun, 06 May 2007 17:14:27 -0600

Dave Wade wrote:

I know there is no commercial value in it, so it won't
happen, but wouldn't it be nice if IBM realeased a
software emulation that worked like the original
XT/370 that emulated both the Hardware and CP calls
and so would allow CMS itself to be run native on
Linux or Windows...
.. oh and of course would license CMS for such
an evironment.

XT/370 was codenamed washington ... stripped down CP kernel running on
modified 68k processor that provided 370 emulation (for problem and
some privileged instructions). The "370" had its own dedicated
processor memory. Running under dos was a program called "cp/88" and
the CP kernel would communicate with "cp/88" for emulation of I/O
operations (i.e. cp/88 provided real device i/o support and
communicated back and forth with the cp kernel).

the original model had 384k "370" memory ... and I did some
application studies which showed that after the fixed cp kernel memory
requirements ...  that cms applications frequently would "page thrash"
in the remaining real memory. Exaserbating the problem was that all
disk i/o (both cp paging and cms file i/o) involved communication with
cp/88 which would then simulate the operations on XT hard disk that
had 110millisecond avg. access.

the publishing of the elapsed time & page thrashing results
resulted in a corporate decision to ship the product with 512k "370"
memory ... which involved a six month schedule slip ... which lots of
people blameed on me.

However, in this time window ... I was allowed to incorporate an
enhanced page replacement algorithm (over and above what i was able to
ship in the vm370 resource manager) ... and CMS "paging access method"
filesystem support ... i.e. page-mapped operation ...  which I had
originally done on cp67/cms ... but never shipped in standard vm370
release.

the problem was that normal CMS operations are highly disk intensive.
DCSS sharing of applications on mainframes were somewhat able to
compensate for some of this (by having programs & applications
already available in real storage because of use by other users).
However, in the xt/370 configuration none of this was applicable ...
there wasn't enuf real storage for such caching ... and since it was a
single user system ... there wasn't any "sharing" use. however, I had
demonstrated avg of 300percent (or better) thruput improvement with
the paged mapped cms filesystem support for disk intensive
operations. The page mapped CMS filesystem support also allowed for
asynchronous operation on program loading ... allowing large block load
of CMS "module" into whatever available real storage ... but also
allowing some asynchronous overlap of CMS application execution with
loading of the program (keeping all the asynchronous activity straight
and hidden from cms by playing games with page invalid/valid
bits). The page mapped CMS filesystem support also had some
enhancements for attempting to do contiguous (physical) allocation
when MODULE was generated (and/or written to disk) ... which could be
subsequently leveraged when program was loaded.

the same adapter board was later made available in ATs and called
AT/370.

the "follow-on" was a full-blown 370 in separate box with 4mbytes of
memory code-nam