Casual Trolling for INfo: Hard Disk SMART temperatures

[BonzaiDuck]

As I've been boldly candid about my activities and priorities, and I'm cleaning up my computer-parts Hiroshima work-room today and this week, my long saga of ducting mods is on hold.

When I first undertook to mod my '94 Compaq ProLiant case, I'd planned to put a duct-box around the RAID-drive cage, but haven't yet got to it.

Today I was monitoring my SMART (internal) hard disk temperatures with my RAID5 array of Seagate 7200.10 drives. Seagate's user manual and specs do not speak to acceptable "internal" temperatures, but rather to the ambient temperatures around the drives, saying that the drives can operate at ambient temperatures up to 69C.

Obviously, since a fan is pulling air across these drives -- even with lackluster CFM -- the ambient is closer to room temperature and between 22 and 28C.

Everest reports SMART temperatures for these drives as between 41 and 43C at idle, and those temps rise to around 47 to 48C when the drives are working hard.

What are good "SMART" temperatures to expect from comparable hard disks? Supposedly the perpendicular technology goes some distance in reducing drive operating temperatures, but I've never seen any study which shows how significant the improvement -- if any -- can be. In the 1990s, Seagate had a reputation for building "toaster-drives," and Western Digital was rumored to be cooler. IBM/Hitachi also had a better reputation for "cool-running" -- with the exception of the Raptors (for good reason).

I migrated from Seagate, to Western Digital, to IBM-and-then-Hitachi -- and now I'm using Seagate again.

Anyone have any insights on this matter of internal hard disk temperatures?

 

[Old Hippie]

Everest reports SMART temperatures for these drives as between 41 and 43C at idle,

I like mine at no higher than that when working.

 

[PCTC2]

My 7200.9 500GB SATA Seagate runs at 33C idle to about 38-40C load. It's standalone for itself, but below it is a 40GB 7200.7 IDE running at 24C-30C and above it is a WD 80GB Caviar IDE @ 36-42C. However, I did have a WD 320GB SATA and 500GB SATA that did run comparably cooler than my 7200.9 by about 4C.

EDIT::: If you're wondering what happened to those WD HD's.... I fried them. I heard somewhere you're not supposed to fill HDD's beyond 70% capacity. Each WD I loaded to around 100% capacity and they died rather quickly. I've, since then, kept capacities around 65% of the total drive.
EDIT:::::: Plus, I never defragged them and they were probably pretty nasty fragmented drives. And I dropped one of them.

 

[ChefJoe]

Perpendicular recording technology wouldn't lead to a temperature decrease with just itself... it allows the drive to run with fewer platters/gb and increased density so it could lead to less mechanical heat for the same amount of use.

 

[BonzaiDuck]

Thanks, folks.

I hadn't paid attention to "SMART" data before, and this is perhaps the first time I've had a hardware controller that makes that data transparent.

There are four drives in the RAID5 cage, and there is about 3/8" of space between them. I know from reading guidelines about HDDs that this is decent clearance, but I can see now that I should attend to improving the airflow.

Apparently, just hanging a 140mm fan an inch behind the cage is not enough to provide adequate airflow, although some cases and disk arrangements I've seen before don't offer any of these airflow opportunities.

 

[BonzaiDuck]

Here's some interesting links I dug up. The second is actually linked within the first, but it was good enough to post separately. In fact, as a PDF, I downloaded it and stored it in my folder of "computer-project documents."


Discussion about HD SMART temps and user experiences

Failure Trends in Large Disk Drive Population

 

[Old Hippie]

Failure Trends in Large Disk Drive Population

I was going to mention this study in my first post. It concludes "higher HD temps are not consistant with higher failure rates".

In my last life, I was a millright for 31 years. Heat was a major factor in mechanical and electrical failure. A HDD is an electro/mechanical device.

To me, cooler is better. And, 'till I see that cooler temps reduce component life, the cooler, the better.

 

[BonzaiDuck]

Oh, I agree. The reason I frequent this forum is an obsession with cooling everything, or keeping everything within a narrower range around an acceptable temperature mean.

But I don't think the authors forged those statistics.

I once had a Maxtor SCSI-2 drive that "locked up" after a year's usage. In those days, my computer-case interiors were warmer than I allow these days. I was rather upset about this, because back then, this was a 250 MB SCSI-2 drive that cost around $500 or more. I remember also buying a Seagate 2GB around that time for just under $2,000!!!

[Please -- don't ask me with exasperation -- "WHAT . . . have you DONE . . . . with your MON-EY!!]

Over a month's time, I took it out of that computer case -- a midtower Enlight. I consulted with a friend who worked for the Navy on submarine tape-recorders. He mentioned that sometimes, if there was an unusual temperature differential between the drive's operating temperature and the room-ambient, that the disk-heads -- when parked upon the system's last shut-down -- would get themselves "vacuum-locked" due to the decrease in temperature to a somewhat lower room-ambient. He recommended holding the drive between the thumb and fingers, and then twisting vigorously sideways back and forth (as though you were mixing a can of paint).

It worked.

This is not inconsistent at all with some of the observations on that web-page, or in the article. So conceivably, you would be better to have a hard disk running between both idle and load in a range of 42C to 45C, than to have a disk running in a range between 35C idle and 45C load, but it may be better in both cases to reduce the spread between room ambient and combined idle and load temperatures. And I'm not even sure of that -- it is . . . . deductive speculation.

Even so -- it's time to start cutting some Lexan rectangles to extend my ducting system to draw air from a better-sealed hard-disk hot-swap cage. If I can increase air-speed over those exterior surfaces of the hard disks, I can probably make the temperatures drop somewhat. I'm just not sure what to expect.

By the way -- I, too -- feel like an "Old Hippie." My friends were "dropping out" in the late '60s, but I kept my nose in the books. Then I graduated to being a Yuppie. So finally, at age 52, I grabbed my retirement checks and dropped out. That's a better way to drop out, I think, than joining a commune in Taos with no income in 1969.

 

[BonzaiDuck]

. . . . Not that my "Chrome Lightning" computer is the center of the universe, nor should it be a center of attention -- except for me.

But here's an update on what some have said in this thread seemed to be "high" SMART temperatures for my RAID5 disks.

Room temperature: 77.5F

Idle HDD 1: 42C
Idle HDD 2: 43C
Idle HDD 3: 43C
Idle HDD 4: 41C

Running Everest Ultimate's Stress test on local hard disks:
[after 20 minutes]
HDD 1: 46C
HDD 2: 46C
HDD 3: 46C
HDD 4: 44C

I'm thinking that I'm using a 3Ware 9650SE PCI-E hardware controller, and that the Everest probably attempts to stress the array as it would any single drive. But I'd have to look at how the test uses the read and write throughput capability on this setup, since the Everest benchmarks correctly bench the array at between 250 MB/s and 120 MB/s depending on the type of "Read" tests. The "Write" tests apparently destroy all data on the disk, so I'll need to find some other way of making the assessment. But if the read performance is similar to THG tests conducted on this card and published within the last year or so, I'd expect the write tests to be similar to theirs.

However limited the Everest stress-test, a 3 to 4C spread doesn't seem to be a "bad" thing. I'd expect the differential to be the same if I can bring down the idle temperature.