There are plenty of guides around detailing how to configure RAID arrays and information about different RAID types.
An issue I have had to deal with for both myself and providing support to other persons is what disk should be used for an array.
This guide does not really apply 100% to large production environments, the Admins of these systems should already know this.
I am aiming this guide at the end user, people who want to setup a RAID array on their home PC/Server or NAS.
The main reason for this guide is that there has been some changes in the hard Drive industry recently.
BRIEF TECH BACKGROUND – Full wiki for ERC
Seagate – ERROR RECOVER CONTROL (ERC),
WD – TIME-LIMIT ERROR RECOVERY (TLER)
SAMSUNG/HITACHI – COMMAND COMPLETION TIME LIMIT (CCTL)
All of these terms/technologies describe a method that can be used to to determine the amount of time a hard drive takes to recover from a read or write error.
On a normal, single non-RAID disk this is not an issue and rarely needs to be changed. The problem in RAID environments is that it must be determined whether the error should be dealt with by the RAID controller or the drive. There can be a conflict where the drive is trying to recover an error, in that time the RAID controller sees it as unusable and normally results in the drive being marked as degraded and therefore requiring a rebuild of the array.
The features mentioned above resolve this issue. Previously even consumer desktop drives had, or could have this feature enabled, either by a series of Linux commands, a firmware flash, or a utility, for eg WD used to have WDTLER.exe which is no longer available.
The issue now is that most new consumer desktop drives do not support these features. The main reason I suspect is that manufacturers want customers to buy their Enterprise or RAID version drives. These drives are more stable and designed for use in RAID arrays, however, as you would expect they’ll cost you more.
Before buying any new HDD that you intend to use for a RAID array, check the model number for compatibility with the relevant technology listed above.
If the feature is not enabled, see if it is possible to enable it using one of the method mentioned. If not then you should find an alternative drive to use.
I would also recommend, budgets permitting, using Enterprise/RAID drive versions for fault tolerance, performance and reliability.
These issues may not effect every setup, it will be hardware and RAID controller dependant. If the drives are not performing huge amounts of read/writes then the error may never occur.
This is only a guide, the responsibility, as always is yours. Always check for system compatibility, whatever components you buy!