Nelson Bartley wrote:
>Question for you guys, why do you dislike WD hard drives?
>
>I've currently got a 20GB 7200 in my server and it runs beautifully.
>It's quiet, I've never had a problem with it, and it's plenty fast. I
>know of all the IBM hard drive problems, however where are these WD
>problems comming from?
>
>NB
>
OK
Firest of all, WD is made about as cheap as possible. It has a single
platter and two heads, which makes the track density about as high as
the technology will allow. Other manufactirers tend to use more
platters and more heads, and sometimes dedicate one head to clocking the
data on the others.
WD reps have stated unequivocally at several shows that they DO NOT
support linux at all, only Windows and Solaris. That tends to indicate
that there is something seriously out of spec with there stuff since
they are unwilling to risk broad scrutiny with highly optimized linux
drivers. It also suggests that they have submitted their own drivers to
those op systems which tend to cover deficiencies in hardware.
My experience with WD drives includes a WD200AB which ran on a RedHat
system at a whopping 1.62Mb/s when set to UDMA66.. I tinkered with the
settings using hdparm and eventually settled on UDMA33 with a
performance increase to over 9M/s.
I use a WD64AA to reproduce disk errors and find workarounds for support
purposes. It does signal service. I can always count on it for
"Seems like memory missing as install crashes"
"hda: lost interrupt"
"0x51 {DriveReadySeekComplete}"
Type errors on a randomly occurring basis. I know if I can tweak the
chipset and the kernel to manage that drive, I can tell the customer
what to do to get running, besides swapping out his horrid disks.
WD has an extended track record of failures and bad drives. I once
called them about a faulty drive, and it was the smoothest sales
experience I EVER had. They obviously do not short their customer
relations department. But then that is spending money to do over the
job that wasn't done properly the first time, and seems to me to be a
scary mismanagement. Build it right the first time,,, An airline tha
has a VP in charge of lost luggage would scare me just as much.
Now, on to facts about testing by Mr. Linux-ide himself.
http://kt.linuxcare.com/kernel-traffic/kt20000214_54.epl#2
Yes, it is not recent, but WD hasn't changed. They still have odd
timing and they still have a very bad workaround for leaving out the
hardware to do 57-byte CRCs and the result is a noise in transmission
from the computer becomes a permanent error on read-back. The blow-off
makes it look like the drive is faster but at the expense of a gamble on
data integrity. Anyone around computers for a while knows that if an
error has nonzero probability, it doesn't matter what the odds against
it are, it WILL occur, and the probability only tells you how often to
expect it. And the hardware/software implementation had better chack
for it and setup a recovery. And WD seems to rely on the low
probability, which is just how Andre Hedrick characterizes it.
And I am still seeing 8 times as many reports of problems with install
or stability from WD owners as from Seagate owners. I rarely see any
from Maxtor owners, and even the slew of reports lately from owners of
recent IBM disks who are running them 24/7 does not match the continured
high level of complaints (about our OS) from WD owners.
Civileme
>
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
