sectors/track IBM 3380 
Author Message
 sectors/track IBM 3380

A question about this showed up in the newsreader where I work.

I believe that the 3380 was IBM's first attempt at a fixed blocksize
disk architecture. I've heard that it writes in bursts of 32 bytes,
then chains them together with an overhead of 12 bytes per block to
simulate a variable blocked architecture.

The maximum track capacity was 47476, with another 12 bytes consumed
by the block overhead. A track capacity of 1484 bursts of 32 bytes
comes out to 47,488, so the arithmetic seems to match this story.

Of course, standard IBM access methods only support blocksizes up to
32760, so the full track could only be used by EXCP coding.

Half track and 1/3 track blocksizes were 23476 and 15476, which seems
to involve another 256 bytes(8 bursts) per block. I don't know if this
also applies to full track writing.



Fri, 02 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:
Bert Manning) writes:

>A question about this showed up in the newsreader where I work.

>I believe that the 3380 was IBM's first attempt at a fixed blocksize
>disk architecture. I've heard that it writes in bursts of 32 bytes,
>then chains them together with an overhead of 12 bytes per block to
>simulate a variable blocked architecture.

>The maximum track capacity was 47476, with another 12 bytes consumed
>by the block overhead. A track capacity of 1484 bursts of 32 bytes
>comes out to 47,488, so the arithmetic seems to match this story.

>Of course, standard IBM access methods only support blocksizes up to
>32760, so the full track could only be used by EXCP coding.

>Half track and 1/3 track blocksizes were 23476 and 15476, which seems
>to involve another 256 bytes(8 bursts) per block. I don't know if this
>also applies to full track writing.

Actually, all of IBMs new DASD hardware is FBA.  The 3390/9345/9395
(RAYMAC)/etc are FBA devices with a controller that makes them look
like ECKD devices externally.  FBA was the best idea to come out of IBM
years ago that never caught on with mainframe users.  I suspect this
was due to lack of MVS support (VM and VSE do support FBA).

--
+----------------------------+
|  Jeffrey C Barnard         |
|  Barnard Software, Inc.    |

+----------------------------+



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:
>A question about this showed up in the newsreader where I work.
>I believe that the 3380 was IBM's first attempt at a fixed blocksize
>disk architecture.

IBM had a previous FBA disk drive. I don't know the drive number though
(3370? 3375?). It wasn't successful because the software didn't support it
very well. It had 512-byte sectors.

+--------------------------------------------+
|              Michael Quinlan               |

|       http://www.primenet.com/~mikeq       |
+--------------------------------------------+



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

[snip]
: I believe that the 3380 was IBM's first attempt at a fixed blocksize
: disk architecture.
[snip]

The 3380 is a CKD device.  Previous FBA DASD were 3370 and, before that,
3330.  Your other info looks correct.

--

-- my opinions are my own, etc., etc.  <== usual disclaimer --



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:


>>A question about this showed up in the newsreader where I work.

>>I believe that the 3380 was IBM's first attempt at a fixed blocksize
>>disk architecture.

>IBM had a previous FBA disk drive. I don't know the drive number though
>(3370? 3375?). It wasn't successful because the software didn't support it
>very well. It had 512-byte sectors.

3370 rings a bell, though you understate the ineffectiveness of the
software support!  The access methods didn't support it at all ....
The following is from memory, but I think it is correct:

The 3380 was the first disk to exceed 32760 bytes per track, but was
otherwise merely a larger and faster 3350 (i.e. count, key and data).
The 3330 was the last of that range that supported sectors, and the
access methods dropped support for the set sector channel command in
DFP (i.e. shortly after the introduction of MVS).

Nick Maclaren,
University of Cambridge Computer Laboratory,
New Museums Site, Pembroke Street, Cambridge CB2 3QG, England.

Tel.:  +44 1223 334761    Fax:  +44 1223 334679



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

: [snip]
: : I believe that the 3380 was IBM's first attempt at a fixed blocksize
: : disk architecture.
: [snip]

: The 3380 is a CKD device.  Previous FBA DASD were 3370 and, before that,
: 3330.  Your other info looks correct.
 ^^^^^^
  I don't think so...  Perhaps the 3344?  The 3330 came in two flavors
model 1 and model 11.  They were definately CKD devices!  :)

: --

: -- my opinions are my own, etc., etc.  <== usual disclaimer --

--

Running MVS, VM and VSE on a Pentium 90 under Linux and Unixware!



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:

> The 3330 was the last of that range that supported sectors, and the
> access methods dropped support for the set sector channel command in
> DFP (i.e. shortly after the introduction of MVS).

Nope, all IBM CKD DASD support the set sector channel command. Just look
at a 3880 or 3990 hardware manual. All channel commands supported are
listed including "set sector". DFP and most modern access methods do not
use set sector anymore since they are written to the relatively recent
"Asynchronous I/O" spec. from IBM. This means that the channel programs
use define extent and locate record, instead of the older "Synchronous
I/O" seek, set sector and search ID channel programs.

If you want to or have old code, the old channel programs still work.  As
far as FBA or CKD, a look at the specs. for the 3380 and up devices you
can see that CKD blocks consume chunks of 512 bytes. So as far as the disk
drive is concerned it's FBA, sort of.  The controller hides everything.
3370's were CKD but not supported by MVS.  3375's were FBA.

--

1226 Starlit Rd.                  |     (714)494-6086
Laguna Beach, CA 92651-3035       | Fax:(714)494-3072



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:


>>A question about this showed up in the newsreader where I work.

>>I believe that the 3380 was IBM's first attempt at a fixed blocksize
>>disk architecture.

>IBM had a previous FBA disk drive. I don't know the drive number though
>(3370? 3375?). It wasn't successful because the software didn't support it
>very well. It had 512-byte sectors.

3310 and 3370 were both FBA devices.  3310 were used only on small systems like
4331s. Their capacity was quite small (<100MB?  too many years ago!).  3370
drives were roughly equivalent to 3350s (~350MB).  If memory serves, FBA
was also used on the ill-fated 9370 series of machines.

FBS was never supported by MVS, only VM (where it was quite successful).  This
lack of MVS support likely contributed to the fact that FBA never caught on.

--
============================================================

Consulting Researcher                   Davis Centre 3334
Computer Systems Group                  UW Extension 4679



Sat, 03 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:



>> The 3330 was the last of that range that supported sectors, and the
>> access methods dropped support for the set sector channel command in
>> DFP (i.e. shortly after the introduction of MVS).
>Nope, all IBM CKD DASD support the set sector channel command.

The "Set Sector" channel command is unrelated to CKD vs. FBA. The Set
Sector command is used to let the channel detach from the device and
service other requests during the rotational delay.

+--------------------------------------------+
|              Michael Quinlan               |

|       http://www.primenet.com/~mikeq       |
+--------------------------------------------+



Sun, 04 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

: : The 3380 is a CKD device.  Previous FBA DASD were 3370 and, before that,
: : 3330.  Your other info looks correct.
:  ^^^^^^
:   I don't think so...  Perhaps the 3344?  The 3330 came in two flavors
: model 1 and model 11.  They were definately CKD devices!  :)

I may have been thinking about the 3310 mentioned in another post on this
thread.  I know there was something in the low 33n0 series that was FBA.  

The 3340 and 3344 were both CKD.  I know about THOSE for sure, since we
used to have a bunch of 'em.  The 3340 had a whopping 65mb of storage (12
tracks per cylinder, 695 cylinders, 8kb per track) on a single removeable
HDA.  

The 3344 had the same basic geometry except that it was fixed and had 4
HDAs, each the same as a 3340.  Physically, they looked just like the 3350,
and even shared many of the same parts.  

Having admitted to actually USING these beasties for something other than a
boat anchor, I guess I've also admitted something about my age.  I might as
well also admit that we stored our program source code and JCL on 80-column
Hollerith punch cards at the time!  Ah, the "good ol' days"...  :-D
--

-- my opinions are my own, etc., etc.  <== usual disclaimer --



Sun, 04 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380
[snip]
: 3370's were CKD but not supported by MVS.  3375's were FBA.
[snip]
actually, it's the other way around.  3370 was FBA, 3375 was CKD, and I
think you're right about it not being supported by MVS.  Wasn't the '75
also the first one to have a "C" model that let you attach 370 channel
cables to both "ends" of a string?

--

-- my opinions are my own, etc., etc.  <== usual disclaimer --



Sun, 04 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380
Maybe this will help.

DEV      BYTES  TRK  BYTES   CYLS  CYL    TRKS  M/BYTE  AVG   ROT  TRANS
TYPE      /TRK /CYL   /CYL    /1M /DEV    /DEV    /DEV  ACC   DEL  RATE
-----                                                   MS    MS    MB
2311     3,625  10  36,250    28   200    2,000      7  ??     ??     ??

2314-A1  7,294  20 145,880     7   200    4,000     29  60   12.5   .663

3310    16,384  11 180,224     6   358    3,938     65  27    9.6  1.031
  (BLKS)    32         352 2,048        126,016

3330-1  13,030  19 247,570     4   404    7,676    100  30    8.4   .806

3330-11 13,030  19 247,570     4   808   15,352    200  30    8.4   .806

3340     8,368  12 100,416    10   348    4,176     34  25   10.12  .885

3344     8,368  12 100,416    10   696    8,352     69  25   10.12  .885

3350    19,069  30 572,070     2   555   16,650    317  30    8.4   .806

3370    31,744  12 380,928     3   750    9,000    285  20   10.1  1.859
  (BLKS)    62         744 2,048        558,000

3375    35,616  12 427,392     3   959   11,508    409  19   10.1  1.859

3380    47,476  15 712,140     2   885   13,275    630  16    8.3  3.000
3380x2  47,476  15 712,140     2  1770   26,550  1,260  16    8.3  3.000

Chris Langford, Cestrian Software



Sun, 04 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:


>[snip]
>: I believe that the 3380 was IBM's first attempt at a fixed blocksize
>: disk architecture.
>[snip]

>The 3380 is a CKD device.  Previous FBA DASD were 3370 and, before that,
>3330.  Your other info looks correct.

   The 2314, 2319, 3330, 3330-II and 3350 were all CKD. I believe the
3340 was the earliest FBA. It ran on a 4361 under DOS/VSE I worked with
once.

Regards,

Alex Vrenios, kx9i

Arizona State



Sun, 04 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:
>actually, it's the other way around.  3370 was FBA, 3375 was CKD, and I
>think you're right about it not being supported by MVS.  Wasn't the '75
>also the first one to have a "C" model that let you attach 370 channel
>cables to both "ends" of a string?

I think you mean the 3350 C models.  It wasn't the channel cables that
went both places, but the CUDI (Control Unit-Device Interface) cables
from the control unit to the DASD.  I don't know for sure if any other devices
offered this, but I think the later DASD offered redundant controller logic
in the head of string so it wasn't necessary.


Mon, 05 Jan 1998 03:00:00 GMT  
 sectors/track IBM 3380

Quote:


>>actually, it's the other way around.  3370 was FBA, 3375 was CKD, and I
>>think you're right about it not being supported by MVS.  Wasn't the '75
>>also the first one to have a "C" model that let you attach 370 channel
>>cables to both "ends" of a string?

>I think you mean the 3350 C models.  It wasn't the channel cables that
>went both places, but the CUDI (Control Unit-Device Interface) cables
>from the control unit to the DASD.  I don't know for sure if any other devices
>offered this, but I think the later DASD offered redundant controller logic
>in the head of string so it wasn't necessary.

Yes, the previous comments were basically correct. The later 3375 units
added a 'tail of string' "D" model for alternate path support.

---more text---

Thanks - Gary



Mon, 05 Jan 1998 03:00:00 GMT  
 
 [ 35 post ]  Go to page: [1] [2] [3]

 Relevant Pages 

1. 3330, 3350 & 3380

2. 3380 Performance

3. 3380 Performance

4. Tracks, Sectors, Heads, cylinders, etc...

5. track/sector data

6. How many tracks and sectors in a disk?

7. RAW Track/Sector reading under Windows 95-98 (real mode interrupt calls)

8. How know the number of sectors/track

9. Tracks, Sectors, Heads, cylinders, etc...

10. Writing at boot sector to the first sector of a disk

11. DOS Sector => Physical Sector

12. Q: How to follow a file sector by sector

 

 
Powered by phpBB® Forum Software