Why do new disks fail?
Often a new VMS system will report a faulty disk in the first few months. It's important to remove faulty disks and replace them promptly.
Sometimes, these new and faulty disks will seem to work when tested later, after removal. So why were they marked as faulty?
The failure rate of a Disk follows what is called a "Bathtub curve" with high failure rates early on due to small defects which were not detected at manufacture, followed by a period of reliable operation, and finally increasing failure rates as the disks wear out.
Bathtub curve [1]
Early Disk issues often are due to very minor defects causing a few bad blocks. Modern disks are good at masking these and reallocating the blocks elsewhere, but this often means retrying the write several times before reallocating the block and writing it to another part of the disk, so the disk appears to run very slowly for a while.
Constant recording systems such as VMS rely on the disk throughput being constantly high, and this reallocation can slow the disk significantly so data is lost. The VMS is usually tolerant of short periods of delay and even of loss, but will mark a slow disk as faulty if it does not rapidly recover and if data is constantly lost. This is important in a RAID array as it allows the working drives to store the data correctly and removes the non-working drive from the active array so no data is lost overall.
Quite often a young faulty drive can be checked by the manufacturer's check utility which will be very patient when it reaches slow areas of disk, and force the reallocation of any bad blocks. The check utility will then declare the disk is fully working. It has been through its problem phase, and providing it is now fully working it can now be put back into the array as a working drive. Tolerate this once, but it misbehaves again, reject it.
The most likely time for a disk to fail is in the first few months of operation: around 5% can fail in the first year [2]. The disk will then become more reliable, and between 1 year and 3 years it will be at its most reliable (around 1% failure rate per year). After 3 to 5 years the risk of failure starts to increase and can reach 11% per year [2], but the drive is still likely to last another 3 years and many disks will work well for 8 years or more before failing.
Summary:
- Promptly remove and replace a drive which has failed, even in a new system.
- Check the drive, and reuse it if it's a fairly new drive which is now working well. Give it one more chance.
- Reject any drives which repeatedly fail.
References:
- [1] Bathtub_curve.jpg: Wyatts derivative work: McSush, Public domain, via Wikimedia Commons
- [2] https://www.backblaze.com/blog/how-long-do-disk-drives-last