[rescue] 36G SCA drives still looking

[Charles Shannon Hendrix]

Fri, 04 Mar 2005 @ 10:41 -0600, erie said:

> The only thing that worries me about SATA drives is that the hda is 
> (quoting a friend of mine that works for a HD mfr) "identical to the 
> ones used in our IDE line"..

Well, some IDE drives are identical to their SCSI counterparts too.  It
depends on a lot of factors like target market, etc.  

The promise of SATA was supposed to be that they would be new/different
drives.  The reality is that some manufacturers cheat: their SATA drives
are IDE units with an IDE->SATA bridge controller.

Seagate Barracudas are real SATA drives.  I've yet to hear much bad news
about them.

On the other hand many (all?) Maxtors and WD drives are fake SATA,
except their "high end" drives like the Raptors.

It helps a lot to avoid the low end "desktop" drives, from everyone.
For example, the Seagate Medalist.  They aren't just slower than the
Barracuda and Elite drives, they are low quality as well.

I personally think most Maxtor drives are junk, based on my own
experienced, and high failure rates reported by merchants (in the form
of returns/replacements) and people building RAIDs.

Western Digital is inconsistent.  Some drives are very good, others seem
to be build specifically to fail.

For example, I have a 4-5 year old Western Digital Caviar.  It has no
errors, has never had to remap a bad sector, and it has nearly 40K hours
on it.

Two other WD drives I had died at the two year mark, and were either the
same model or the same family.

Inconsistent quality control?

Fujitsu IDE I don't know anything about.  I have three Fujitsu UW SCSI
drives and nothing exciting to report, except that they seem to have
higher non-media failure rates than my IBM and Seagate SCSI drives.

I assume that's SCSI command errors?

IBM/Hitachi desktop (Deskstar) drives seem to have high failure rates,
but the Ultrastars seem to be very good.  I have not seen an IDE version
of the Ultrastar.

I have 4 IBM Ultrastar SCSI drives, all 5 years old, and as far as I can
tell they have no errors or other problems. The pair in my Sun server
have been running hot that entire time, usually hovering between 50C and
60C.  Sun Aurora cases don't keep the drives very cool.

> anyone notice how long ata drives last these days? I've got 10+year
> old st-506 drives on the shelf that will fire up still, a few of them
> after 4+ years of 24/7 operation.

I retired a pair of Seagate Hawks after 6-7 years of 24/7 operation,
because one of the spindle motors started vibrating.  No errors on
either drive though.

> I have a bunch of SCSI drives in an array on a proliant 6000 that ran
> 24/7 for 2+ years before I got them over 2 years ago, anyone know of
> an IDE drive that will?

Yes, some of them will.

I don't like any of the Maxtors, and other similar consumer level
drives.

I don't know anything about Maxtor and WD "enterprise" drives.

I'm spooked to try them.

I have good luck with Seagate Barracuda IDE drives.

One thing to keep in mind too is that most drive failures and data loss
occurs because people never check to see if their drives are giving any
warning signs.

For example, checking /var/log/* on UNIX systems, or the event log in
Some drives last longer than others.  Also keep in mind that some IDE
drives are identical to their SCSI counterparts.  It depends on the
target market of the drive, and other things.  For example, some of the
Seagate Barracuda IDE line are SCSI drives with an IDE board on them
instead.

Otherwise, yes you have to be careful.  For example, a friend's father
used to work for Quantum, and he said that they built IDE drives from
parts rejected by quality control for their enterprise lines.

A tech from Segate told me that their Medalist drives were made from
components not fit for the Barracuda and Elite lines.

Most desktop users don't push drives hard, and are highly price
sensitive.

Seagate seems to be doing pretty well, and they offer five year
warranties on their Barracudas.  I'm not worried about drive failure
with the two Seagate SATA drives I bought.

I wouldn't touch a Maxtor, even though a friend of mine says they have
improved.  I've never had anything but trouble with them.

Hitachi's desktop line has a bad reputation.

Western Digital is hit and miss.  Some drives seem to last a long time,
others die quickly.  I have a WD Caviar that is going on 4-5 years, no
remapping at all, and shows no signs of trouble.

Another WD drive I got died in less than 2 years.

In most cases when I've helped people recover from driver failure, their
system had been warning them of impending drive failure for weeks or
even months.

Most drives sold these days will report their errors via S.M.A.R.T.

Of course, I find it frustrating that all of the SMART data is not
perfectly standard, but hopefully that is improving.

For some reason, smartctl cannot get the status of my SATA drives.  I
don't know if it is because they show up as SCSI drives or what.

> reliability is suspect, and until I see some real world numbers (not 
> MTBF #'s, I worked at DEC when we rolled out our 5 1/4" drives, I know 
> how MTBF is calculated..)

One thing I've never understood is why there are no public empirical
tests done on drives.

It really wouldn't cost that much money, certainly less than some other
consumer advocacy costs.


-- 
shannon "AT" widomaker.com -- ["All of us get lost in the darkness,
dreamers turn to look at the stars" -- Rush ]