home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Simtel MSDOS 1992 December
/
simtel1292_SIMTEL_1292_Walnut_Creek.iso
/
msdos
/
info
/
dtarev1.arc
/
DTAREV1.ASC
Wrap
Text File
|
1989-01-18
|
53KB
|
1,321 lines
Disk Technician Advanced
a review By Mike Focke
15 January 1989
I wrote an earlier (June 1988 CPCUG Monitor, SPINRITE.ARC on
a Bulletin Board) review of SpinRite(tm), a product that
tests your hard disk, refreshes the address markings and data
on your hard disk and optimizes the placement of sectors on
your drive. When I saw an ad for a new version of Disk
Technician Advanced(tm), a similar product, I thought it
would be interesting to do a comparison.
Disk Technician Advanced is one of three versions of Disk
Technician marketed by Prime Solutions Incorporated. Over
150,000 copies of various versions of Disk Technician have
been sold. Version 3 Revision 5.22 was tested.
1. Disk Technician is copy protected, works with up to two
physical MFM drives of up to 32 megabytes and lists for
$99.95. It allows you to run tests on your hard drives
and floppies and to repair any repairable sectors found to
need repair. Destructive low-level formatting is provided
for all machine types. Only AT type machines will allow
use of the non-destructive formatting capability, not PC,
XT or PS/2 types.
2. Disk Technician+ is copy protected and works with large
capacity drives as well as both MFM and RLL controllers.
It lists for $129.95
3. Disk Technician Advanced (hereafter "DTA") is not copy
protected and provides additional features including non-
destructive formatting on all machine types (except PS/2
types with translating controllers) and interleave
optimization. DTA lists for $189.95 and can be found
below $100.
To run DTA, you copy the DTA floppy to a disk with DOS
installed. You reboot your system from that newly created
floppy to insure that no TSRs, FastOpens, or caches are in
use. You are then asked to run an initial monthly run which
allows DTA to establish a data base with information on any
errors discovered. Non-destructive low-level reformatting is
done if sector interleave optimization is found necessary on
this initial monthly run.
Once the data base has been established, Daily, Weekly and
Monthly runs are provided to maintain the data base and to
allow the "Artificial Intelligence" of this version of Disk
Technician to determine:
If there is significant deterioration of the drive or
controller such that you should be warned of the
probability of future problems.
If a particular "Soft 1" read/write error (one that could
be corrected by the controller without DOS ever seeing an
error) is happening so frequently that the cluster in
which the sector occurs should be marked as "bad" and
your precious data moved to a safer place.
TESTS
DTA run against a previously interleave-optimized Seagate
ST225 20 megabyte drive took:
Monthly Run - 3 hours and 41 minutes
Weekly Run - 1 hour and 23 minutes
Daily Run - 4 minutes
After the initial run of both SpinRite and DTA, I created a
"bad" spot on the ST225 by over-tightening the bolts that
attach the side of the drive to the computer's case and doing
a single sector write. This over-tightening is known to
cause a distortion in the drive's case, so the heads will
write data slightly askew as a result. After I wrote the
"bad" data, I relaxed the bolts' tension so that the heads
would now align correctly but the data would still be "off
center".
Both SpinRite and DTA detected this sector as "bad" and both
choose to restore it to use after reformatting the track and
rewriting the "off center" data. On runs after the initial
one, DTA would reassuringly report that it was monitoring
that previously "bad" spot on the disk.
DTA uses a cluster of your drive, which it marks off as "bad"
in the FAT so it can not be erased, to store a safety
identification number. It can then be sure that the data base
it is using and the drive it is reading match. Despite DTA
showing that none of the clusters on the drive are "bad", you
will have a cluster that other utilities will show as "marked
bad".
A NEC 20 megabyte drive had never shown any "bad" spots and
neither SpinRite nor DTA detected any "bad" spots despite
repeated runs.
The true test is not one where the drive is already known
to be good, I needed a drive with some marginal sectors to
test. Larry Babcock (A-Quality Personal Computers in Vienna,
VA) volunteered 6 known bad hard drives that had been removed
from computers when customers upgraded. One, a Tulin TL226
20 megabyte drive, was just what I was looking for, a working
drive with some "bad" spots marked on the case. I used the
Western Digital Controller's BIOS logic via DOS's DEBUG to
low-level format the Tulin and marked off only one of the two
tracks marked "bad" on the case. I used DOS's FDISK to
partition the drive and DOS's FORMAT to high-level format the
drive. I knew that DOS's FORMAT program would catch some of
the "bad" spots in the remaining "bad" track and mark those
clusters as "bad" in the File Allocation Table (FAT). Once I
had done this, I had a drive with examples of the two ways
that sectors can be marked "bad". There were 30,720 unusable
bytes in 15 "bad" clusters of 4 sectors of 512 bytes each, a
track containing 17 sectors and 8704 bytes was marked "bad
track" and the rest of the sectors on the disk were good
according to DOS's FORMAT command.
I ran SpinRite's Extremely Thorough Pattern Testing mode, the
one which tests each sector 84 times. SpinRite reacted to a
series of consecutive "bad" sectors by issuing a message that
might have convinced a novice that there was something
so seriously wrong with the drive that it should be replaced.
There wasn't, and I allowed SpinRite to continue. SpinRite's
testing allowed it to return 11 clusters to use. SpinRite
found 6 clusters that were "bad" that DOS's FORMAT did not
find. SpinRite marked 19 clusters "bad" in the FAT.
DTA's initial run took over 5 hours and detected "Soft 1"
errors in three sectors but repeated testing showed that they
were so infrequent that they could be ignored. DTA's initial
run detected "Soft 4" errors in 2 sectors. "Soft 4" errors
are soft errors that occurred even after the track had been
re-low-level formatted by DTA and repaired. These "Soft 4"
errors were cause for marking 2 clusters "bad" in the FAT.
DTA's initial run detected 17 hard errors that were never
readable and so those 17 clusters were also marked as "bad"
in the FAT. In short, DTA had detected the same 19 errors
as SpinRite and 3 errors that SpinRite had not detected.
For all of these drives, SpinRite suggested an optimum
interleave of 3 giving a transfer rate of 171kb/sec. DTA
choose a more conservative interleave of 4 giving 128kb/sec.
According to DTA's Tech Support, DTA sacrifices speed to
reduce the probability of "soft" errors.
In theory, if SpinRite and DTA were both equally effective in
detecting errors, DTA should have found no additional errors
beyond the ones SpinRite found. Both were better than DOS's
FORMAT.
RELIABILITY DESIGN
There are two ways a program can request the BIOS to do a
low-level format of the hard drive to prepare the surface
to be partitioned and high-level formatted in preparation for
DOS file storage. The two ways are to use a "low-level
format the entire drive" command (Interrupt 13 Function 07),
or to repeatedly use the "low-level format track x of
cylinder y" command (Interrupt 13 Function 05). Using the
track by track approach, a program could read data into
memory, format the track, and then write out the data back to
the disk. This method is known as "non-destructive" since it
does not destroy your data. Using the entire disk method,
you can not store the data from the entire disk. This method
is called a "destructive" low-level format.
The DTA manual warns that nondestructive low-level formatting
of PC/XT systems "will require many thousands of seeks which
may cause less than optimal formatting on the one hand, and
needless mechanical wear-and-tear on the other". I have seen
similar comments in the documentation to a shareware product
called HDTEST (HDTST121.ARC). The author of HDTEST, Jim
Bracking, says that "the reason for this is that the XT hard
disk controller performs a recallibrate before executing the
format track command.... a seek to cylinder 0 is performed
before each format track command." AT/286/386 systems do not
do a recallibrate when doing track-by-track low-leveling.
Prime Solutions recommends a destructive low-level format be
used on PC/XT systems but there is no discussion of this
concern when the default settings for DTA are described on
page 20 of the manual. If it is such a concern, why did
Prime Solutions choose "use non-destructive format" as the
default option and give no on screen warnings?
While DTA's default is to non-destructively reformat only a
drive detected to be running with an unoptimized interleave
and to selectively non-destructively format an area of the
disk where errors have been detected, SpinRite non-
destructively reformats as its default in all but its surface
scan mode.
Repeated use of SpinRite on multiple XTs causes me to wonder
about the seriousness of this concern. After all,
reformatting is not done every day even with SpinRite. And
even HDTEST, whose previous versions prevented non-
destructive formatting for PC/XT class machines, now allows
PC/XT non-destructive formats.
When I asked, Prime Solutions' Tech Support reminded me that
the drive was doing 2 seeks per head per cylinder or over
2,400 seeks on an 4 head 615 track 20 megabyte drive (ST225).
That doesn't seem like so many compared to the number I do in
a normal day. Prime Solutions' Tech Support says that low-
level formatting using the "format entire disk" command is
more precise than low-level formatting with a series of
"format track n" commands. I have no way of proving or
disproving this.
-
It's important to understand how a disk controller and DOS
combine to read your data because SpinRite and DTA make
important and different assumptions about what to use to
signal the existence of an error.
In simplistic terms, a program issues a read command to DOS
which issues a "read a sector" command to the BIOS which
issues a "read a sector" command to the hard disk BIOS. The
Hard Disk BIOS issues a "read a sector" request to the
controller. The controller talks to the drive through the
hard disk electronics. The hard disk electronics position
the head over the appropriate cylinder, select the required
head and begin to read information from the track into the
drive electronics. When the requested sector is detected (by
the address markings passing under the head), the data
portion of the sector is read. Statistically, there are a
fairly large number of times where an attempt to read a
sector results in a read that is detected by the controller
to be bad. The data is good on the disk but the drive or
electronics somehow did not get a good read. This is called
a "soft" error. A "hard" error is one where the data is
actually damaged on the disk.
At this point, the controller itself initiates a retry.
Peter Norton, in his book "Hard Disk Companion", says the
IBM-XT controller does 10 reads before reporting an error.
If the sector is successfully read in any of these tries, the
sector is returned by the controller to DOS with no
indication that there was any difficulty.
Only when the controller determines that retries alone will
not successfully read the sector, and only when consecutive
reads result in the same syndrome, does the controller use
Error Correction Code logic to try to recover the data which
caused the read error. A data error can occur in from 8 to
25 consecutive bits and still be recovered, depending on the
controller. If the ECC is able to mathematically reconstruct
the data, the controller returns the data to DOS as being
read successfully, and tells DOS that ECC was used to read
the data via a status that says "data ECC corrected"
(Interrupt 13 Function 01 Status 11h).
It would even be possible for DOS to use the "data ECC
corrected" status to know how to remap your data to a safe
section of the drive. Neither IBM or Microsoft are doing all
that they could to protect your data!
If the data is not successfully recovered by the 10
controller retries or by the ECC logic, the controller
returns an error code to the BIOS which, when passed to DOS,
will cause DOS to retry this whole process up to three times.
Only after all these 30 attempts by the controller, the ECC
and DOS, do you get the "I/O Error, Abort Retry, Ignore?"
message. This may explain why retrying a read so seldom
results in a successful read.
If DTA asks to read a sector and controller retries and ECC
fail, DTA goes into a "seek profile" which attempts to
position the heads so that the sector can be read.
Sometimes, this can take hours in an attempt to recover your
data. Even if these efforts fail, DTA displays an error
message showing you what file was damaged. DTA writes a
sector of all blanks to show you what data is missing and to
allow you do know how to recover.
SpinRite, when it encounters that same hard error, reads the
data with a "read long" command (Interrupt 13 Function 0Ah)
which reads the sector complete with ECC. The ECC is then
corrected and the data (complete with "bad" bits) is written
out to a safe sector. Even SpinRite's manual says that its
approach is to save all the data it can rather than give up
just because a few bits have been changed! It is this that
Jerry Pournelle is referring to in his article in Volume 11
Issue 1 of January 2, 1989 in InfoWorld. My guess is that
this was discovered by someone running a Virus Protector that
CRCed the executable files. When, after running SpinRite,
the test of the CRC of an executable file did not compare
correctly with the CRC stored from a previous Virus
Protection run, the file was compared and the data shown to
be changed by SpinRite.
SpinRite configures itself to the controller and BIOS to
detect when the ECC had to be used to detect that an error
occurred. It traps the status before DOS has a chance to
throw it away. SpinRite is designed with the assumption
that, if the controller repeatedly reads the sector without
ECC having to be used, the sector is surely readable when DOS
allows the use of ECC. And if the ECC had to be used by
SpinRite, the sector should be avoided.
SpinRite, in all but its Surface Test mode, reads a track
worth of your data into memory and then does a low-level
track reformat. SpinRite then begins write/read tests (up
to 84 of them). These tests are all done immediately after
the address markings are refreshed and without any
intervening seeks to throw alignment off. If the tests done
to the sector in this series of tests are all error free (if
Error Correction Code logic is never invoked by the
controller), the sector is deemed usable. In all but the
Extremely Thorough Pattern Testing (84 write/read cycles),
SpinRite only tests the sectors already marked
good to make sure they are still good.
In the Extremely Thorough Pattern Testing mode, SpinRite
tests even the sectors marked "bad" (on the track by a low-
level format or in the FAT by some other utility) to see if
they can be restored to use. No matter that the sector has a
history of problems. SpinRite can't know about that history
because it maintains no historical data base. No matter that
sectors just in front of, in back of, and to the side of this
one have errors, SpinRite's logic tests just one sector at a
time. No matter if the initial attempt to read your data had
to be retried 9 times by the controller, SpinRite doesn't
know about that because it depends on ECC being invoked to
determine if there is a problem. If the cluster passes the
tests, it is marked as available for data storage. Even if
the cluster was previously marked "bad" in the FAT because
the last pass of SpinRite found errors, the cluster will
still be restored to use in SpinRite's Extremely Thorough
Pattern Testing mode. This "restore to use because it tests
ok today" approach seems directly counter to the reason you
run such a utility. You don't run it to restore marginal
areas to availability. You run it to map out any doubtful
sectors. And if there was a problem with that sector last
month, I'd feel better if SpinRite just left the spot marked
"bad".
DTA times the length of time it takes the controller to
return the data and status to the controller. It is designed
with the assumption that, if the time is too long, the
controller must be rereading the sector. To DTA, it does not
matter if the data is successfully read, what DTA learns is
that there was a problem reading the data. By knowing what
sectors produce such errors and with what frequency, DTA can
determine what spots on the drive surface are predictable
"trouble spots" and are to be avoided.
DTA does not do a low-level format before it begins Daily,
Weekly, and Monthly tests. Thus the write/read and read
attempts are done with the address markings in their typical
state and not as they would be after a fresh reformat.
Unlike SpinRite's tests, there are many seeks between the
initial format and today's DTA tests, so that, if there are
any alignment problems, they can be detected.
DTA looks for a case where the time to read takes too long,
showing that the controller had to perform a retry. DTA
calls this a "Soft 1" error indicating that the controller
had to retry one or more times but the data was eventually
read successfully and without the use of ECC. If DTA shows
such an error, the sector producing the delay, and other
areas around that sector, are analyzed to see if there is a
pattern. The history file is examined to see if there is a
pattern. "Soft 1" errors are normal, but too many of them
too frequently indicate problems with that spot, even if the
spot could be written to and read from successfully many
successive times. Perhaps the magnetism weakens with time,
or perhaps the head was positioned over this spot and a bit
of magnetism was generated on power up or power down. Unlike
SpinRite, DTA's default is not to restore an area to use just
because a freshly formatted sector can be successfully
written and read without ECC being necessary. DTA follows
the philosophy that an area which produces too many "Soft 1"
errors, even if they could be recovered by repeated
controller retries, is probably not the area to which you
want your data written. Since these "Soft 1" errors occur
normally in every drive and controller combination, it is
DTA's touted "Artificial Intelligence" logic which determines
how many is too many.
The combination of detecting "Soft 1" errors by using timing
to detect that the controller is having trouble reading a
sector, the use of a history file and some logic to determine
when too many of them occur too frequently, and the refusal
to restore a sector found "bad" by any previous test, seems
like more conservative logic than that used by SpinRite.
-
SpinRite offers four levels of test (Surface Test, Minimal
Pattern Testing Depth, Average Pattern Testing Depth and
Extremely Thorough Pattern Testing). You pick how long you
are willing to spend in running the test according to how
much reassurance you want.
DTA provides you four levels too (Initial, Daily, Weekly
and Monthly). There is no discussion in the manual of what
should happen if you don't adhere to that schedule. A phone
call to Prime Solutions reveals that the only result of not
running according to the recommended schedule will be a data
base that is less effective and a longer time between
verifications that your data is readable.
-
Only in its Extremely Thorough Pattern Testing mode will
SpinRite attempt to restore to use a cluster that had
previously been marked on the track or in the FAT as "bad"
but which now tests as good. The second time you run
SpinRite, it will mark as "good" clusters which it had found
in its first run to be "bad" unless you change SpinRite's
mode of operation from its default.
DTA will normally attempt to restore to use any clusters that
can be successfully repaired only when doing a non-
destructive low-level format run to optimize interleave, only
when the track itself was marked "bad" by some low-level
formatter which accepted input of a set of manufacturer
supplied bad tracks, and then only when the clusters pass
exhaustive tests. DTA never marks as good a cluster which
was marked "bad" in the FAT by some previous error detection
program.
It makes sense to use this "restore to use" option once and
only once. The manufacturer does exhaustive tests on a hard
disk's surfaces and provides you a list of tracks they
consider as having potential for errors. The manufacturer
specifies a track as "bad" because they can't tell you the
sector or cluster since the manufacturer doesn't know what
type of controller or what version of DOS you will be using.
(Different controllers (MFM, ARLL, RRL) write different (17,
26, 31) numbers of sectors on the track. So, an error that
might be in sector 8 on a MFM controlled drive might be in
sector 15 on the same drive if it were ARLL controlled. No
version of DOS maintains a FAT (where "bad" sectors are
listed) where the units are sectors. All DOS versions mark
space in units of a cluster. The size of the cluster depends
on the version of DOS and the size of the drive.
No controller's low-level formatting allows you to mark off
"bad" spots at the sector or cluster level. When you follow
the manufacturer's recommendation, you must mark the whole
track as "bad", even when only one sector is "bad". The only
way you can tell which cluster within the track is "bad" is
by testing with something like SpinRite or DTA. The first
time you run the non-destructive low-level formatting option,
let SpinRite or DTA restore the "good" sectors from a "bad"
track. But the second time you run SpinRite or DTA, you do
not want the sector that tested "bad" on your first run
restored to use, even if it tests "good" today.
You are likely to use SpinRite in a way that will restore
previously "bad" clusters to availability, because the
testing depth that common sense suggests you use, the
Extremely Thorough Pattern Testing mode, is the one that will
mark the "bad" clusters that test good this time as available
for use. This is a serious flaw with SpinRite.
Because DTA provides Monthly, Weekly and Daily runs which
don't restore to use previously detected "bad" sectors, you
are much less likely to use DTA in a mode that restores "bad"
clusters to use than you are SpinRite.
-
SpinRite, except for its surface scan mode, non-
destructively formats your drive each time it is run. This
rewrites the address portion of your data, refreshing the
magnetism of the address bits.
DTA, except during its initial run (and then only if sector
interleave optimization is found necessary), uses a non-
destructive format "only in those specific areas which
actually need repair".
DTA seems to think that if it is good, don't mess with it.
If it is bad, then reformat it and test it. Here is an area
where the procedure DTA uses seems to contradict the reason
you run one of these programs - prevention. DTA refreshes
the address markings only when an error is detected and not
to prevent a future error by periodically strengthening the
magnetism in the critical address markings.
-
When I replaced my "D:" drive half way through these tests, I
needed to "RESET" DTA's data base so that the history of
errors would apply to the new drive and not the old one.
Unfortunately, when DTA does a data base "RESET", it removes
the history information for all of the installed drives. So
the history of the "C:" drive on which I had used DTA for a
month was lost too.
USER INTERFACE
Both SpinRite and DTA can be used against two physical drives
in a single machine. With SpinRite, if there are two drives,
you are given a menu showing the two physical drives and are
asked to choose against which you wish to run SpinRite.
DTA's drive selection is from the command line or a more
cumbersome on-screen option override. There were no
instructions in the manual on how to use DTA on two drives
in the same machine. Since one of DTA's advertised features
is the creation of a historical data base, I needed to know
if the data base on my DTA floppy could safely contain data
about two drives or if I should run from a different floppy
and have a different data base for each drive. I had to call
Prime Solutions' Tech Support to find out that DTA will
safely combine the information about 2 physical and 26
logical drives into the one database. It turn out that there
are no instructions in the DTA manual for installing your
XT's clock accessing software on the DTA floppy.
Consequently, when I boot the DTA floppy and DTA presents a
date of 1/1/1980, my first instinct is to change the date.
Changing the date puts me in manual option entry mode. Had I
just accepted the date or had I been running from an AT which
provides the correct date without additional clock accessing
software, I would have been in a mode that would have run the
appropriate tests against both drives. DTA should be changed
to allow date change to have no effect on further option
selection. Prime Solutions has told me that date and time
entry will not put you in manual entry mode in future
revisions. The programmer has called me twice to clarify the
approach to be taken.
-
Urgent warning messages are less visible with DTA than
with SpinRite.
Warnings in the SpinRite manual are bordered in bold black
and are prominently displayed. SpinRite issues warnings as
the only item of text on a screen and with strong clear
language.
DTA's warnings are buried in the manual and are in smaller
type than the normal text. DTA on-screen warnings are buried
among other text on a very busy screen.
-
SpinRite gives clear on-screen warnings against running TSRs,
FastOpen and caches. SpinRite even tests for the presence of
cache and refuses to run.
DTA relies on warnings buried in the manual or on-screen
warnings that are by no means as clearly displayed or as
explicit as those of SpinRite.
-
The DTA manual suggests that it is best to run DTA from a
floppy "because it stores..information which it can use to
restore access to your hard disk in case of a crash". There
is no further explanation about this "recovery" feature in
the manual. A call to Prime Solutions' Tech Support revealed
that it is the partition sector that is saved and can be used
if the partition sector on your hard drive is damaged when
DTA begins its run.
-
SpinRite uses the conventional "ESCape" to abort.
DTA tells you to use the "Control and Break" keys to abort
the program. Not every keyboard has a clearly marked "Break"
key! Mine doesn't.
-
Even the first time it is run, SpinRite estimates the time it
will take to complete and refines that estimate as the
reformatting is being done.
No on screen estimate of time remaining or percent completed
is available with DTA. DTA refers you to the manual for time
estimates but the manual contains no table of function and
disk size versus time. When asked, Prime Solutions says that
it is difficult for them to estimate how long the process
will take since DTA make changes in the testing modes due to
the number and types of errors DTA encounters during a run.
-
SpinRite is interruptable and allows you control of your
computer when the current track has been safely put back on
the disk. The next time you run SpinRite, it asks you if you
want to restart from the point of interruption.
DTA is interruptable, but the manual says you may have to
wait as long as 20 minutes after you ask before you get
control of your machine from DTA. DTA is not restartable at
the point of interruption. It is more difficult to design a
way to stop a DTA run and restart it later than it is with
SpinRite. With SpinRite, you only have to store off the
number of the next track to be tested. With DTA, you would
have to store off the accumulated variables that have been
built with DTA's logic that analyzes the error patterns. DTA
is working on restart for a future release.
-
Both SpinRite and DTA can be run from batch mode with options
selected by command line arguments and from interactive mode
with options selected via typein.
SpinRite allows you to make option selections by moving a
highlite selection bar over the desired option and pressing
"Enter".
DTA lets you choose options via a command line interface or
override the options DTA suggests by hitting "ESCape" and
then typing in the name of the option you desire. DTA would
do well to copy SpinRite's separation of option selection
from performance statistics display.
-
SpinRite does some hardware testing to determine that your
system is working well enough for SpinRite to safely reformat
your drive. The purpose and result of each test is clearly
shown on the screen.
DTA's manual does not say DTA does similar tests before
beginning the process of reading and writing on your disk
drive. Conversations with DTA's Tech Support confirm that
DTA does the tests but prefers to keep the details hidden in
an attempt to seem un-technical.
-
Both SpinRite and DTA provide performance statistics.
SpinRite produces its statistics as a byproduct of its
hardware testing before beginning its low-level formatting.
SpinRite shows both access and data transfer speeds.
With DTA, the normal screen shows areas where the results
of the tests might be displayed, but you must explicitly
request the test, and the mechanics of making the request are
buried on page 30 of a 71 page manual. DTA only displays
access speed and not data transfer speed.
-
Every SpinRite option choice you are asked to make has an
on screen explanation as you move the option selection
highlite bar over the option. Even the explanations have
page references to text within the 34 page manual if you
want to know more about what is happening. But you really
don't need the manual.
To figure out how to pick each option in DTA, I had to read
and reread the DTA manual.
-
DTA's manual has a note that it is an in-house printing and
they offer to send you the professionally printed manual when
it is available. There's nothing wrong with the print
quality. It's the quality of organization within DTA and the
corresponding difficulty in clearly describing the interface
in a manual that is at fault. While there is an index, such
needed entries as "interrupt operation" and "time estimates"
were not in the index.
-
Each product has a screen saver mode that allows a small
block of information to be constantly moved around the screen
so that an image does not become burned into your monitor's
screen. This is necessary because runs against large drives
can take many hours.
The SpinRite screen blanker shows the percentage of test
still remaining. I'd like the time remaining until
completion too.
With DTA, you have to return from screen saver mode to the
main display screen to see your progress. Even then, you are
shown only the cylinder and head currently being processed,
and not a time until completion.
-
There are warnings in the SpinRite documentation and on its
screens against running SpinRite without first removing any
copy protected software from the disk. Some programs use a
form of copy protection that deliberately creates a false
"error" on your drive. If SpinRite were to correct the
error, your program could think you did not legally install
the program and could refuse to load.
There is no such warning in the DTA documentation. Prime
Solutions says there doesn't need to be a warning. DTA's
approach is to never restore to "available status" a sector
that was detected as having an I/O error. So a copy
protection "error" would be detected as an error, would not
be moved and would be marked "bad".
-
SpinRite doesn't make any changes to your AUTOEXEC.BAT file.
DTA's QUIKSTALL(tm) modifies your AUTOEXEC.BAT file without a
backup being made. SPA.COM, DTA's timed head parking
program, is added to your AUTOEXEC.BAT as the first line.
Running SPA as the first thing in my AUTOEXEC.BAT somehow
caused later lines in the AUTOEXEC.BAT to abort due to a lack
of environment space. When SPA was run as the last line in
the AUTOEXEC.BAT, no ill effects were noted. DTA's Tech
Support had never heard of this happening in anyone else's
machine. When sent a copy of my AUTOEXEC.BAT file, they
determined that this was a well known effect of SPA being a
Terminate and Stay Resident (TSR) program which will use
environment space. That, in combination with my extremely
long PATH and PROMPT statements, created the problem. I must
say that their Tech Support was very interested in the
problem and will know how to help the next person who reports
it.
-
Both SpinRite and DTA can produce printouts or files of the
results of the testing.
SpinRite will allow you to write results of a single run to a
file which you would then have to manually rename if you
wanted it available for comparison with a future run.
DTA allows the results of many runs to be stored in a history
file for later review and comparison with previous runs. DTA
can also print out a "bad sector" listing showing the
Cylinder, Track and Sector of all detected errors.
-
SpinRite will allow you to select an interleave if your
experience is different from SpinRite's estimation.
DTA will only reinterleave at the interleave that it
suggests. It was impossible for me to tell DTA that I knew
from a year's experience that the drives in my machine would
run reliably at an interleave of 3. DTA insisted it knew
best and interleaved them at 4 costing me about 26% in
transfer speed.
ADDITIONAL UTILITIES
Both SpinRite and DTA come with head parking programs.
SpinRite's PARK just parks the heads of all installed drives
on command.
DTA's QUIKSTALL will mark off a cylinder on the drive as
"bad" in the FAT to allow a safe parking area.
DTA's SPA will move the heads on all installed drives over
this safe area after a user selected time of between 1 and 15
seconds of hard disk inactivity. SPA command line options
will park on command and turn off the timed park.
LICENSING
SpinRite offers several licensing options. Option 1 allows
use on any machine you personally own. Option 2 allows use
of SpinRite on all computers owned by a company beyond the
initial machine for a fee of $25 per machine supported.
Option 3 allows use by consultants on other people's machines
for a fee of $10 for each use. Option 4 allows Systems
Houses to use SpinRite as part of their setup and test
procedure for systems they build, provided they include
SpinRite promotional literature with each system they sell.
Option 4 is free of additional cost.
DTA is licensed only for use on a single machine.
SUPPORT
I called Gibson Research several times and never did get
through. I also sent a copy of the review and received no
comments.
I called Prime Solutions probably six times with questions.
Each time I was helped and several times ended up talking to
the principle author. A copy of the review drew an hour
conversation that they paid for.
SUMMARY
If you are going to buy now, I'll make it easy.
If you want a reasonably priced product with a beautiful user
interface and fine documentation, a product that you're
likely to use occasionally, one that will both refresh all
the bits of address and data on your hard disk and conduct
tests to mark off problem sectors that are detected in the
current test (but which should be run in its "do not restore
good clusters to active use" mode)...then buy SpinRite. But
wait for the next version which promises fixes to some of
SpinRite's error detection deficiencies.
DTA is a much expensive product. Your drive will transfer
data slower if you use DTA than if you use SpinRite. DTA has
a less pleasant interface and a more complex manual than
SpinRite. But DTA uses a method of error detection that is
superior to SpinRite's. And DTA's use of a data base of
previous test results to determine when to mark off a sector
as "bad" has been found in my tests and in those of InfoWorld
(July 25, 1988) to be superior to SpinRite in detecting and
marking off bad sectors. For doing what you buy such a
program for, protecting your data from being written on a
marginal spot on your drive, DTA is clearly superior. DTA
will get even better in future versions as their improvements
to the user interface are implemented. If I could just buy
one, I'd spend the extra money and buy Disk Technician
Advanced on sale.
FOR FREE
Here is a free technique for doing essentially what SpinRite
and DTA do for you. Warning, it takes a lot more work
though.
About once every six months:
Run CHKDSK /F to insure the logical integrity of your file
system.
Backup your hard disk using your backup software.
Do a low-level format with differing interleaves and test
using SpinTest (a free performance tester available on
BBSs as SPINTEST.ARC) until you have determined the
optimum interleave.
Low-level with the optimum interleave, DOS's FDISK and
High Level DOS Format.
There is a freeware program DISKTEST distributed via local
Bulletin Boards as DTST14.ARC which does tests of the time
it takes to read each of the sectors on your drive. It
reports a list of which took excessively long to be read
and which of those are not already marked as "bad" in your
FAT. It suggests the use of Norton's Utilities to find
what file is stored in the slow sectors that are not
already marked as "bad" in the FAT, to move the file to
another place on the drive, and to mark the slow sector as
"bad" in the FAT. Not as easy as DTA but it does give you
a view of how DTA might be using as a test the length of
time it takes to read a sector. And it is free.
Mark those sectors as "bad" in the FAT using FM32 from
FM32.ARC available on local Bulletin Boards. FM32 is
interactive and will display the current state of your
FAT, will allow you to make changes, and will allow you to
write the FAT back to the drive if you are satisfied with
your changes.
Rerun DISKTEST and mark any sectors still found slow and
repeat this cycle until you have several consecutive
DISKTESTs which show no slow sectors that are not already
marked "bad".
Re-Format using the DOS Format command's /s and /v options
to put the system on the drive and write a Volume Name.
Reload your files by doing a file restore using your
backup software.
Run Bracking's DISKTEST once a month to reassure yourself
that no new spots have become marginal.
If you are willing to do this, you will be doing more than
90% of the people who use hard disks are doing to refresh the
address markings and test for "bad" sectors. And you'll be
safer than you are by doing nothing. You can see from this
description how much easier running SpinRite or even Disk
Technician Advanced would be.
WARNING
No matter which technique or program you use, make sure that
you buy a good backup program first. Back up your hard drive
before you use either SpinRite or DTA for the first time.
Not because they are buggy, but because there is no way
either company could test for all the unusual hardware
combinations we have in our computers.
_____________________________________________________________
PRODUCTS MENTIONED
___________________________________________________________
Disk Technician Advanced Prime Solutions Inc
$189.95 list 1940 Garnet Avenue
San Diego, CA 92109
Phone: 619-274-5000
SpinRite Gibson Research Corp.
$59 list PO Box 6024
Irvine, CA 92716
Phone: 714-854-1820
SPINTEST.ARC Gibson Research Corp.
free see above
HDTST121.ARC Jim Bracking
$35 registration 967 Pinewood Dr
San Jose, CA 95129
Phone: 408-257-0945
DSTS14.ARC Samual H. Smith
free Sysop
The Tool Shop
BBS Phone: 602-279-2673
FM32.ARC Thomas Stepka
Free 7702 Lunceford Lane
Falls Church, Va 22143
Norton's Utilities Peter Norton Computing Inc
$100 2210 Wilshire Blvd
also #186
Norton's Utilities Santa Monica, CA 90402
Advanced Edition
$150 Phone: 213-319-2000
A-Quality Computers
138 Church St NE, Suite F
Vienna, VA 22180
Phone: 703-938-8665
_____________________________________________________________
AUTHOR'S BIOGRAPHY
Mike Focke is a Manager for Honeywell Federal Systems Inc in
McLean, Va. He has 22 years experience in computers
including 6 in micro-computers. For the Capital PC Users
Group's Monitor, he has previously reviewed XT and AT hard
disk controllers and hard disk reliability and optimizing
software. He can be reached on the Hard Disk conference of
the Jeff Morley's Interconnect BBS at 703-827-5762.
_____________________________________________________________
HISTORY
12-10-88: Initial Draft
12-18-88: Considerable Revision following comments from Alan
Dowd relating to SpinRite
12-19-88: Clarifications following phone conversation with
Prime Solutions' Tech Support. The review was also
sent to Steve Gibson for comment. I have an unusual
attitude for a reviewer, it is more important to
get it right than to get it published. So I
frequently allow the manufacturer to review the
article for factual accuracy. The opinions remain
my responsibility and they are changed only in
response to clearly demonstrated fact.
12-28-88 Reorganization for Clarity of elements dealing with
function and those dealing with interface.
12-29-88 Additional info concerning the I/O process following
additional conversations with DTA's Tech Support.
1-4-89 Info on the SPA and environment space conflict
included.
1-7-89 Tests on the marginal drive added. Comments on ECC
added.
1-9-89 SpinRite's treatment of non-ECC correctable data
added.
1-15-89 Added info about difference in interleave between
SpinRite and DTA. Added info about loss of hard disk
history for both drives even if only one is
replaced.
Queried Prime Solutions about Winn Rosch's very
negative DTA review in the Jan 9, 1989 PC Week.
They replied that they believed that he ran DTA on a
machine which had been partially reinterleaved by
SpinRite and that the different approaches the
monthly and daily tests take was the cause for the
different results. I was disappointed that Winn did
not take the time to retest following their
suggestion. Perhaps the pressure of doing 5 reviews
a week got in the way of doing it right.
First BBS version released
