Hi, Jay:
As far as I know:
IBM 2311 DASD had 3625 bytes per track, with 10 tracks per cylinder
IBM 2314 DASD had 7284 bytes per track, with 20 tracks per cylinder
IBM 3330 DASD had 13030 bytes per track, with 20 tracks per cylinder
IBM 3350 DASD had 19069 bytes per track, with 15 tracks per cylinder
I believe the use of the "portable" blocksizes,
RECFM=FB,LRECL=80,BLKSIZE=3120 for source files, and
RECFM=U,BLKSIZE=6144 for PDS load module libraries, came into widespread
use with the introduction of the MVS IPO (Installation Productivity
Option) -- IBM began distributing a pre-configured MVS system that could
be easily installed and "up and running" quickly at new customer sites.
This was introduced somewhere around 1978 to 1980, if I remember
correctly. The idea was that the datasets on the IPO all had these
block sizes, so they could be installed on most any popular DASD devices
then available.
This also became important of your worked for an ISV, and had to
distribute your software on tape in a format that could be easily copied
(restored) to disk at a customer site, with a minimum of fuss and bother.
And, to address comments by Rich, responding to comments by Christian,
as I recall, no operating systems then extant had yet come up with "a
better way" to deal with "disk geometry" (as it was called). So, it was
not only IBM operating systems that seem "primitive" by today's standards.
AFAIK, it was not until FBA disks became popular, and in the 1980s, with
the emergence of the SCSI standards, when companies began to figure out
that it was much easier to just deal with a disk unit as a sequence of
addressable blocks, and let the hardware (controller / device) figure
out how to map that onto the actual drive "geometry". Also, it was not
until the advent of SCSI that we really saw the emergence of the idea
that devices (e.g. disk drives) could "identify" themselves, and tell
the controller (or computer) about the drive size, etc.
Actually, IBM invented much of this concept with the IBM System/38,
which was designed starting around 1975, and that was first announced in
1978, but did not see widespread installations until 1980. With the
System/38, all disks were formatted as "pages" of uniform size, and all
"files" etc. reside in a vast virtual address space called "single level
storage" where the objects themselves would actually be spread across
multiple drives to gain the advantage of multiple heads and arms, to
speed access to various parts of the objects. When you add a new disk
drive to the System/38 (or the AS/400 follow-on product), when you power
on the system, it detects the new devices, interrogates them to
determine their capacity, etc., and then "auto-configures" them, so they
are ready for use. So, the big change with the System/38, versus the IBM
mainframes and System/34, System/36, etc., was that you never had to do
a "SYSGEN" (or CNFIGSSP) to configure new devices -- you just plugged
them in, and turned the system on. As far as I know, this was truly the
first version of what would later be called "Plug-And-Play" (by
Microsoft, circa the time of the introduction of Windows 95.)
All the best,
Mark S. Waterbury
On 8/4/2013 10:14 PM, Jay Jaeger wrote:
Not sure of the disk drives 100%, but the numbers 7294, 13030 and
19069 ring bells for me as track sizes - I am guessing for the 2314,
3330 and 3350, respectively. (And a quick check on Google after I
wrote the previous sentence adds support for that guess).
Though I was still involved in mainframe from time to time after 1989,
my exposure diminished after that. I fondly remember one time needing
to plug in some channel cables for a new Intel/Jupiter TCP/IP
interface into a System 390 right around 1989. Intel/Jupiter wouldn't
do it without a bunch of money to fly someone in, and IBM wouldn't do
it, period. Fortunately the IBM SE (who had also been involved in
preparing IBM's bid for that contract -- and lost) came over to me,
and knowing my pretty extensive hardware/electronics background told
me that the FE's were going to have to quiesce the system for 15
minutes that very afternoon (I think for microcode patches) and if I
knew how, then maybe it could be done then (wink wink), and said
something about making sure I didn't strip the screw by first rotating
backwards a turn before screwing it in (wink wink). The SE is still
with IBM, I think. She's a class act and a GREAT person.
That system (which was at Wisconsin DOT) went away shortly thereafter
when the state's computing facilities (and staff) underwent another
level of merger, and I stayed at WisDOT to support VAXen, PC's, Apollo
HP and UNIX, and then web servers and then web application servers.
Jay
On 6/28/2013 12:26 PM, Rich Alderson wrote:
From: Christian Corti
Sent: Friday, June 28, 2013 1:22 AM
PS:
IMHO IBM mainframe operating systems are
completely braindead...
why does the user has to know anything about cylinders and drive
types?
Because that's the way your grandfather did it, and if it was good
enough for him, it's good enough for you! ;->
I stopped being a programmer on IBM systems in 1984, so why do I
still have the numbers 7294, 13030, and 19069 embedded in my brain?
What do they represent? (I do know.)
Rich
Rich Alderson
Vintage Computing Sr. Systems Engineer
Vulcan, Inc.
505 5th Avenue S, Suite 900
Seattle, WA 98104
mailto:RichA at
vulcan.com
mailto:RichA at
LivingComputerMuseum.org
http://www.LivingComputerMuseum.org/