is it the end of the line for SCSI Terminators?by Zsolt Kerekes,
editor
See also:- (parallel)
SCSI SSDs, What
is SCSI?, Storage
History |
When I first started
using SCSI devices as a systems integrator in the mid 1980s the answer to - how
many drives you could actually hang onto an inhouse made ribbon cable? -
seemed to depend on trial and error.
In those days everything had
little mechanical switches, called DIP switches, which enabled you to set up
important configuration info such as the SCSI address and whether or not the
terminators were to be connected on this device or not. The rule of thumb was
that you had to connect the first and last devices physically looped on the
chain of cables. Of course that helpful information failed to provide a simple
solution to the artistic patterns of daisy chains intermixed with star offshoots
around the back of the cabinet which our ingenious wiring people sometimes came
up with. It was hard to get them to accept that sometimes a single long loop
worked better than a lot of little limpet like add-ons.
When working on
a group of so-called "identical" systems I would sometimes be
mystified by an intermittent fault in one storage system, even when all the
indivdual parts tested as OK. On closer inspection I might discover that
some drives had the terminator resistor packs plugged in, while their
counterparts in their twin systems didn't. The culprit would often be another
project, which in the middle of the night cured their own problems and got
closer to customer shipment by "borrowing" components from our own
system, which they knew was still a long way in advance of the critical path.
A
week or so later, I might be handed a packet with a bunch of terminators and
told "These are to return the ones we borrowed." I learned to keep a
secret stash of these little suckers for this type of emergency.
Eventually
we learned of the existence of SCSI analyzers and other similar
test equipment, which was
plug and play, instead of the plug and fall off variety when we tried to see
what what going on by hooking up about 30 test clips from a logic analyzer. That
woudl produce insights only until another project pinched the analyzer, and
deleted the custom trace settings... What do engineers do for entertainment
nowadays? - I wonder.
Now things are a lot easier with almost
everything controlled by software. Smart active terminators can measure the
characteristics of the cabling and reflections on the line on reset and
dynamically adjust the termination. With serial connections like
Serial Attached SCSI
and Serial ATA, the
problems of data skew, cross talk and reflection are greatly simplified by the
simpler cabing design. But proper termination still has a role in reliable
system integration, even if its tacky details are invisible to the user. |
| . |
 |
above - a SCSI cable and accessories ad.
This was the last parallel SCSI cable ad which ran on
StorageSearch.com and dates from
2003. |
| . |
|
|
| . |
| Unlike traditional SSD
designs - in adaptive R/W the ECC/ DSP strength, duration of the write
program pulse and even the virtual block size can all be varied to optimize
the SSD's headline objectives (such as speed or power or usable to raw
capacity) and reconcile them with the flash memory's actual health
condition. |
| Adaptive flash
care management & DSP IP in SSDs | | |
| . |
 | |
|
