BurnInTest V4.0 Disk test questions

[WinEmAll]

I work with a company which uses the Passmark Burnin test V4.0 Pro on various sized Maxtor hard drives. On some tests, after about as quickly as 1 hour, we can see a large difference in the amount of cycles completed on the hard drives. There is no correlation between how big the hard drives are and how many cycles are completed, and whether or not they were formatted in FAT32 or NTFS. It almost appears as if some just randomly go slower. The difference is not small either. An example is this computer next to me which has been running for 26 hours, and has 232 cycles for one hard drive, and 1564 cycles for another. Neither have Critical or Serious warnings either.

On another note, which would you reccomend is better for testing hard drives, NTFS or FAT32 formatting. Also, do you think 48 hours is way too much stress for hard drives? Possibly even permanently damaging? Thanks in advance.

[David (PassMark)]

The disk cycle period is not directly related to the amount of time that has passed. A completed cycle represents the writing and reading of a test file to and from the disk. The size of the test file, the speed of the drive, the amount of background activity, the duty cycle, and the test pattern selected, determine how quickly a cycle is completed.

The size of the test file is in turn determined by the user by entering in the file size (as a % of the disk size) in the BurnIntest preferences window. The duty cycle is also adjustable by the user. So it is normal that the cycle time will vary from one drive to the next, especially as you have stated that the drives are of various different sizes.

As for the file format. I would use NTFS. Why? Only because it is the format in use by the majority of PCs sold today (ignoring floppy disks which still use FAT).

Concerning the test time. Samsung claims a mean time between failure (MTBF) of 500,000 hours. Fujitsu claims and MTBF of 300,000 hours. The are power on hours, rather than hours of heavy load and are probably a exaggeration for marketing purposes. But none the less, 48 hours should only be a very small percentage of the drives overall life.

Or looking at it another way. It is better that your new drives fails while under warranty and still empty or is it better that it fails a month after you have installed it when it is full of data?

------
David
PassMark

[Infinitus]

Desired load and hours for testing a disk

Hard disk drive: WD 1TB RE3
Software version: 5.3.1020 x64
OS: XP x64 SP2

I did a disk test at 100%, but after six hours the drive and the utility both started giving various errors (also seen in System event log). This is supposed to be a good drive with very low failure rates, and now I'm wondering if I stressed it a bit too much thereby damaging it.

In the past couple of years, based on research studies, we have come to know that the MTBF of various disks is basically a bogus number and doesn't say much about a drive's reliability.

The posts above in this thread were from 2004. Please tell me if your recommendation for testing a drive for 48 continuous hours (in random data with random seeks mode) still stands, or should a drive not be tested for so long? For how many hours and under what load would you burn test your own new drive (at most)? Are there any lessons that have been learnt in this regard?

Thank you.

[David (PassMark)]

If you are using V5.3 of the software, you should probably post in the V5 forum, not the V4 forum.

You should also update to the latest V5 release, which is build 1034 at the moment for 64bit. You are 14 releases behind the current release and there have been changes to the disk test in release 5.3 build 1026 that might effect your results.

If a new drive doesn't can't last 48 hours under load, you should return it as faulty. The same drive might experience similar amount of seeking load if it was used for a company database server or similar application.

Regardless of if the MTBF failure numbers are bogus, it shouldn't stop you holding the manufacturers to account.