Write speed for Seagate 1.5G HDD (ST31500341AS) @ 13MB/s?

[sfsilicon]

I was having problems with my Seagate 1.5G HDD (ST31500341AS, FW:CC1H) where the file system kept on getting corrupt. After running Seatools I was surprised to see that it passed, so I put it back into my PC and started copying data to it which is when I noticed that the write speed was very slow (13.1 MB/s).

I did more testing and compared it to the Samsung 2TB drive I also had in my PC and got the below results. For all the testing I copied to and from an OCZ SSD on a different controller to ensure maximum bandwidth. I also updated all the controller drivers to the latest. The Seagate starts in the 100MB/s (shows that the Windows and drive DRAM caching are working), but since I copied +20GB of data eventually the writes dropped down to 13MB/s.

samsung
- read 114 MB/s (16GB of data)
- write 95MB/s (27.4GB)

seagate
- read 110 MB/s (16GB of data)
- write (13.1 GB) (27.4GB)

Is this normal? Should I return it for a warranty exchange? Could someone with the same Seagate drive do a similar test, so I have another reference on a different PC. I have a second drive and it seems slow too, maybe a sign that the drive isn't working fully at spec and may fail in the future. Any help is greatly appreciated.

 

[war9200]

Show smartctl -a /dev/sda (or /dev/sdb) etc output from a KNOPPIX or System Recue CD.

 

[sfsilicon]

The seatools also show smart info. What specifically should I be looking for?

I ran Seatools on the drive and it passed Short/Long DST and Short/Long Generic. It should be good to go, but the performance is really unacceptable at 13MB/s writes. I'm just trying to get some correlation to see if this is normal with this drive and if not I will contact Seagate support.

--------------- SeaTools for Windows v1.2.0.4 ---------------
1/17/2011 9:00:18 PM
Unit Serial: --------
Unit Model: ST31500341AS
Model: ST31500341AS
Serial Number:
Firmware Revision: CC1H
Short DST - Started 1/17/2011 9:00:18 PM
Short DST - Pass 1/17/2011 9:18:05 PM
Long Generic - Started 1/17/2011 9:40:07 PM
Long Generic - Pass 1/18/2011 1:53:14 AM
Identify - Started 1/18/2011 7:10:22 AM
Long DST - Started 1/18/2011 7:22:23 AM
Long DST - Pass 1/18/2011 11:47:12 AM

--------------- SeaTools for Windows v1.2.0.4 ---------------
1/17/2011 5:14:53 PM
Unit Serial:
Unit Model: ST31500341AS
Model: ST31500341AS
Serial Number:
Firmware Revision: CC1H
Test Unavailable - Test Unavailable 1/17/2011 5:14:53 PM
Test Unavailable - Test Unavailable 1/17/2011 5:14:58 PM
Identify - Started 1/17/2011 5:15:06 PM
Short DST - Started 1/17/2011 5:16:16 PM
Short DST Aborted 1/17/2011 5:39:35 PM
Long Generic - Started 1/17/2011 5:39:57 PM
Long Generic Aborted 1/17/2011 8:38:36 PM
Short Generic - Started 1/17/2011 8:39:02 PM
Short Generic - Pass 1/17/2011 8:58:48 PM

 

[sfsilicon]

Ran Crystal Disk Info and attached the screenshot. I have 2 identical seagate drives and both are showing "CAUTION" for reallocated sector count. I haven't test the other drive for write performance yet. Will do that next. Should I RMA both drives? Would you agree that the write performance of 13MB/s is not acceptable?

 

[houkouonchi]

so it looks like 20 re-allocated sectors? I would verify that the volume is not nearly full or writing a file that is heavy in fragments (which could reduce the speed) before doing an RMA. If it has 20 re-allocated sectors that is a bit on the high side and reduced sequential read/write performance can be a sign that drive is dieing but usually I see both reduced write and read performance caused by disk health issues.

 

[war9200]

so 20 in HEX is 32 re-allocated sectors, if its under warranty and you can't get the write speed problem fixed and the fs is getting corrupted, your disk is probably on its way out, can you run a short test (5 min) and then a long test (2-5hr) on the drive and post the results?

download smartmontools for windows then
smartctl -a /dev/sda
or
/dev/sdb, or /dev/sdc, whichever the disk is

smartctl -t short /dev/sda # wait 5 min
smartctl -t long /dev/sda # wait 5 hr

then smartctl -a /dev/sda

post result

 

[drescherjm]

Will do that next. Should I RMA both drives?

Reallocated sectors are not a problem unless the number is growing. Having said that 7200.10 and 7200.11 drives, I can say are less reliable than drives from other manufacturers. And this comes from using them at work.

Would you agree that the write performance of 13MB/s is not acceptable?

This depends on what you are copying and how fragmented your disk is. 13MB/s would be good if you were copying a million 1K files but horrible if you were copying a single 100GB file.

 

[sfsilicon]

Thanks for all the suggestions and feedback. This is really helping me. One quick questions before I go into responding to you. Are there any known cases of people selling fake Seagate drives? The problem drive is an OEM bought from a US retailer. I just noticed the label looks a bit different from the other 2 I have bought. The reason why I am asking is that the warranty check tool of Seagate can't find the drive's warranty status while it can find the other 2 I have. I've contacted Seagate about the problem.

so it looks like 20 re-allocated sectors? I would verify that the volume is not nearly full or writing a file that is heavy in fragments (which could reduce the speed) before doing an RMA. If it has 20 re-allocated sectors that is a bit on the high side and reduced sequential read/write performance can be a sign that drive is dieing but usually I see both reduced write and read performance caused by disk health issues.

The volume was not empty, around 50 percent full when I started. I had formatted it earlier and in one go had copied around 700GB worth of HD files to it, so I'm assuming it isn't heavily fragmented. I then copied 27GB HD files (6 files) from an SSD to get a fairly constant stream of data. The only thing I might have done wrong is that I turned off the write caching a while back because I was getting file system corruption on 2 the 3 drives I had. Both would periodically require a scandisk when Windows 7 booted up. Then it would be ok for a while then all of a sudden a directory would be inaccessible. This normally would happen when I fill the drive more than half full. I got tired of that and worried for my data and copied everything on to a new 2TB drive. During that copy that is when I noticed the slow performance since it took forever to copy of the drive.

I hooked up my 3rd ST31500341AS and it's sustained rate is in the 43MB/s, my 2nd ST31500341AS (which hasn't given me a problem so far) is copying at around 90 MB/s. I also measured and older 500GB seagate drive and it would do 30MB/s writes. All the same test of copying the 6 HD files (27GB) from an SSD. So the 13MB/s on the problem drive is very slow. I will redo the experiment and format the drive and do the copy again.

so 20 in HEX is 32 re-allocated sectors, if its under warranty and you can't get the write speed problem fixed and the fs is getting corrupted, your disk is probably on its way out, can you run a short test (5 min) and then a long test (2-5hr) on the drive and post the results?

download smartmontools for windows then
smartctl -a /dev/sda
or
/dev/sdb, or /dev/sdc, whichever the disk is

smartctl -t short /dev/sda # wait 5 min
smartctl -t long /dev/sda # wait 5 hr

then smartctl -a /dev/sda

post result

I am getting File System errors on 2 drives (one with 13MB/s and one with 43MB/s vs the 90MB/s on the only drive not giving me problems, which is why I am inclined to RMA the 2 problem drives. All three drives were bought less than 2 years ago, so they are in warranty.

The main reason why I haven't done that yet and why I am still doing testing is that Seagate requires you to run the Seatools suite and I have passed both the short (5-10m) and long (5+ hours) DST/Generic tests (see 1st post). The long Generic test includes attempts to fix defective blocks. I recall on the 13MB/s drive it the short tests would hang till I did the long generic test which also took a long time, but after it passed the short test passed as well. I'm guessing it fixed something. Still the 13MB/s performance is unbearably slow, with the threat that the drive might fail soon. I use these drives as storage drives and it takes a long time to fill them at that speed, then if I lose the data it would be a disaster.

war9200 if you think I should still run the short and long test please let me know.

Reallocated sectors are not a problem unless the number is growing. Having said that 7200.10 and 7200.11 drives, I can say are less reliable than drives from other manufacturers. And this comes from using them at work.

This depends on what you are copying and how fragmented your disk is. 13MB/s would be good if you were copying a million 1K files but horrible if you were copying a single 100GB file.

I haven't been monitoring the number so I really don't know. These are being used as back-up for my movie collection. I have read the feedback on the 7200.11 on NewEgg and the failure rate doesn't look like something I would like to trust my collection on.

I think I am doing it correctly. I am copying 6 files which are between 4-6 GB. Since the bandwidth drops so quickly I am typically still copying the 1st file when the caching can't buffer the writes anymore. I see it goes from 200MB/s down to 13MB/s within a minute or two. I'm using an SSD as the source drive to avoid any bottlenecks on that side.

 

[drescherjm]

Try the copy again and then look at the reallocated sectors. Most likely the reason for the slowness is the drive reallocated a sector during the copy. This can take up to a minute or more for each reallocation.

 

[sfsilicon]

Try the copy again and then look at the reallocated sectors. Most likely the reason for the slowness is the drive reallocated a sector during the copy. This can take up to a minute or more for each reallocation.

When I do the copy and see the continued slowness, I should check to see if the reallocation count has gone up?

Is there anyway to fix this problem? Is this an indication that the drive will eventually fail since it is continuously allocating sectors?

 

[drescherjm]

When I do the copy and see the continued slowness, I should check to see if the reallocation count has gone up?

Yes.

Is there anyway to fix this problem?

Do a full format. If possible (assuming you have enough space to move the current data to a different drive). That will force the drive to reallocate any bad sectors it finds.

Is this an indication that the drive will eventually fail since it is continuously allocating sectors?

It definitely could indicate that. At work where I have 100s of drives I have some drives that will have 50 reallocated sectors for years. I have seen others where the count went from 14 to 2000 and then to failure in less than a week. I generally monitor this on my servers and if I see a drive increasing the reallocations each day or each week I swap it out. Then test and RMA if necessary.