[email protected] (Quasar Chunawala) writes:
> I work as an application programmer with a leading bank on CICS/Cobol for
> the past 4 years. Whilst I know, that data on the mainframe is stored on
> disks and tapes, I have never walked in to a data-center. At any mainframe
> data-center, what are different storage media used? Do they still use Tapes
> or 3390 DASD? I did do some research on the Internet, but found that the
> last 3390 was manufactured by IBM in 1993. What *storage media* is used to
> store petabytes of data and information? I also heard the term *DASD Arrays,
> * but I am not quite sure what they are.
>
> I have yet another question. Can DASDs or tapes be virtualized?
in effect, much of all IBM I/O is "virtualized" these days
as you found out real CKD DASD hasn't existed for decades it is all
simulated ("virtualized") on fixed-block disks. past posts mentioning
CKD & FBA disks
http://www.garlic.com/~lynn/submain.html#dasd
FICON channels are a protocol layer (simulation/virtualization) on
fibre-channel standard (FICON layer happens to drastically cut the
throughput of native FCS). past posts mentioning FICON
http://www.garlic.com/~lynn/submisc.html#ficon
disk array was invented by somebody in IBM San Jose disk division
(patent in 1978)
http://en.wikipedia.org/wiki/RAID
but IBM doesn't even have a disk division anymore. disclaimer ... they
use to let me play disk engineer in bldgs. 14&15 where Ken worked
http://www.garlic.com/~lynn/subtopic.html#disk
correctable redunancy codes have been used to *correct* failing bits in
storage. it is also used in forward-error-correcting (FEC) code in data
transmission ... not just parity bits for error detection ... but enough
additional information to correct error transmissions). Common
transmission FEC codes are Viterbi and Reed-Solomon.
http://en.wikipedia.org/wiki/Error_detection_and_correction
disclaimer ... I used to have a project called high-speed data transport
... one of the people working on it was graduate student of Reed and had
done a lot of work on Reed-Solomon.
http://en.wikipedia.org/wiki/Reed%E2%80%93Solomon_error_correction
HSDT also did some work with Cyclotomics up in Berkeley that produced
Reed-Solomon products ... as well as a lot of the work in the cdrom
standard (cyclotomics was later bought by Kodak, one of the founders of
cyclotomics was Berlekamp)
http://www.garlic.com/~lynn/subnetwork.html#hsdt
common disk failure mode is whole disk failure. simplist disk array is
mirroring ... replicate writting every piece of data to two different
real disks. If one fails, there is still redundant copy.
high-performance computing combines striping and error correcting ...
say 32+8 (32 data disks, 8 error correcting disks). A record is divided
into 32 parts ... and 8 error correcting pieces are calculated ... then
all 40 pieces are written in parallel across all 40 disks. It logical
works as a single disk that is 32 times larger than a "normal" disk with
data-transmission happening 32 times faster. DataVault, 32+8 shows up in
1985 for high-performance computing.
http://en.wikipedia.org/wiki/Thinking_Machines_Corporation
Trivia ... Brewster Kahle ... from above, leaves and forms WAIS, Inc
... that is then bought by AOL ... Kahle then creates the wayback
machine.
http://archive.org/index.php
a more complex operation is done for DBMS operation. a 5+1 ... involves
five data disks plus one error correcting disk .... but can read
multiple different records concurrently ... but writes are more complex
the raid wiki article goes into more detail.
--
virtualization experience starting Jan1968, online at home since Mar1970
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
