Can a HDD with a few bad sectors slow down the whole computer significantly?

[Cubytus]

Hello there,

after a hard drive scan with Drive Genius, it found out 29 bad sectors on the hard drive. As DG explained, once they're spotted, they can be permanently disabled in the drive's controller.

Therefore, my primary question is, can these failed blocks slow down the whole computer significantly, as an Apple Genius said when I brought it back for a quick reinstall after a one-time RAM failure? Like, if a large file needs to be written, it will necessarily fragment around these blocks¿

Background: I have bought the drive in 2011, had to return it because of heating issues, and just learned that they never received the original one, and that I consequently lost all remaining warranty. Western Digital never made me aware of that fact, and needless to say, I am extremely pissed that a 5-year-waranty drive 1. would require two replacements in 2 years and 2- that I would have to buy a new one with a ridiculously low amount of "loyalty discount".

 

[ColdCase]

All drives have bad sectors, usually transparent to the OS. Theoretically bad sectors can impact performance but usually not, more often overall capacity. 29 sectors is about 15KB or 120KB, hardly noticeable on GB drives.

 

[TheBSDGuy]

I use Scannerz (http://scsc-online.com/Scannerz.html) to do drive testing. One of the nice features about it is that in addition to detecting bad sectors it can detect timing irregularities.

When a controller detects a bad sector it maps the bad one out and assigns it to a spare sector. Normally when data is written to a drive it attempts to write as much sequentially as possible to a series of sequential blocks on a drive. This way the drive heads don't have to be hopping all over the disk that much. When one of the sectors is remapped it means that instead of following a linear track, the heads must jump to the out of sequence tracks to access the remapped block.

Does this really affect performance? In my opinion, unless you're using a really old drive with a small cache, no. Old drives were a lot slower. I took an old drive that had a bad sector on it, zeroed the drive (which is one way to force remapping) and retested it with Scannerz. Any time Scannerz came to that block it detected a minor irregularity. It read it but it indicates it's slower to read than the rest of the data on the drive.

That was with an old, and I mean OLD (like 13 years) drive. Todays drives are a lot faster and they have much bigger caches. Once the data is in the RAM cache, it's read directly from the drive's cache and is really only limited by the speed of the interface. If it's read only rarely it might not get cached. Overall, your talking about delays on newer drives on the order of microseconds or milliseconds. If delays are much longer, then it's likely a weak sector which may not get remapped by a controller. It all depends on the controller's design and firmware settings.

With that said, I'd be more worried about the fact that you developed bad sectors. It's not abnormal for them to have a few, but I'd test it more frequently for a while to make sure the problem isn't getting worse. It's not uncommon for drive problems to start getting worse and worse once they're detected.

 

[Cubytus]

Well, I suspect these ones appeared after the drive was installed. The corresponding SMART item was downgraded, and Drive Genius provided some additional details and, hopefully, excluded them. I noticed there was something wrong when a large file couldn't be copied elsewhere as it couldn't be read (Error -36 in the Finder).

I am not too concerned about a lack of space as this takes place on a 500GB drive and apparently performance shouldn't be a significant issue, but would like to know why and how would bad sectors appear on a drive that young, as you point out. There's also no way to know how well the firmware performs in this drive AFAIK.

 

[whitedragon101]

can these failed blocks slow down the whole computer significantly, as an Apple Genius said when I brought it back for a quick reinstall after a one-time RAM failure? Like, if a large file needs to be written, it will necessarily fragment around these blocks¿

Yep.

Whenever I have a dying drive that develops bad sectors the sign to check its smart data is massive slowdown. The HD is always reading and writing to the swap file in the background and bad sectors often cause problems such as beach-balling, freezes and system hangs while it finds the data it is after or finds a place to write to.

This just happened with both my HD in my Macbook Pro 17" (early 2011) and then with my girlfriends drive in her 2009 15" MBP. The same cloned image on a new drive fixed the problem as it always does.

 

[ZVH]

The trouble with relying on SMART status is that it's not reliably implemented from vendor to vendor, and how each vendor interprets SMART status and deals with problems also varies. IMHO you actually NEED something to scan a drive once in a while (I do mine w/Scannerz once a month) because SMART status will never mark a bad block as bad if it's unaware of it. In other words, if your current drive is 500GB in size and you're using only the first 150GB and a crash occurs at say 250GB, SMART would be unaware of it because there's really no reason for the heads to be moving over that region to get data. SMART needs to detect an error during a read/write operation, and in some versions I think it only detects problems during a write operation. It's far less than perfect. Critical write ups exist about it in both Wikipedia and a Google study on drive longevity.

Finally, I have an old system (PPC PowerBook) that suffered a head crash some time ago. I split the drive partition into three sections: 2 good, big partitions and one bad partition that had the crashed data in it. After partitioning I just deleted the small partition with the head crash in it. SMART status has been telling me this drive is about to fail any minute now - for almost 5 years!!

 

[TheBSDGuy]

Just for anyone interested:


Wikipedia on SMART:

http://en.wikipedia.org/wiki/S.M.A.R.T.

Google test summary:

http://static.googleusercontent.com...ch.google.com/en/us/archive/disk_failures.pdf

 

[laurihoefs]

Just for anyone interested:


Wikipedia on SMART:

http://en.wikipedia.org/wiki/S.M.A.R.T.

Google test summary:

http://static.googleusercontent.com...ch.google.com/en/us/archive/disk_failures.pdf

These are very good reads indeed. From the Wikipedia article the S.M.A.R.T. attributes highlighted with red are especially interesting, these are also attributes that drive manufacturers consider critical (always an RMA). S.M.A.R.T. has some parameters that don't indicate anything critical, or are vendor specific, but scan errors or reallocations are not such. They are always serious signs.

From the Google study, page 7:

After the first scan error, drives are 39 times more likely to fail within 60 days than drives without scan errors.

Healthy drives have no bad sectors. When reallocations or scan errors start to appear, it is a sign of a failing drive. Statistically these drives fail fairly quickly, but there are always exceptions, like the ones reported on this very thread.

From my experience systems with failing hard drives often exhibit freezing or hanging, I've seen it happen in Windows, BSD and Linux. I wouldn't expect OS X to behave differently.

So my answer to the OPs question, whether failing drives cause slowdowns: yes.