In previous posts, I’ve introduced the concept of RAID and how it can help to safeguard your data. There can be one “fly in the ointment” related to a RAID implementation. That problem is the time that is needed to recover your data from a “crash”. Depending on the size of the drive arrays involved, the recovery process can take many hours.
(I can hear some of you say: “As long as I get my data back, I don’t care if it takes a number of hours”. If that’s true in your situation, then a RAID-and-backup implementation should give you peace of mind.)
If you cannot afford to wait hours to have access to data on a damaged array, there is still hope. Some servers support what are called “hot swappable” drives. These are drives that can be taken out and replaced (I.e. swapped) while the server is turned on.
In a hot swappable implementation of a RAID, a damaged drive can be swapped with an empty replacement drive without turning off the server. The server will detect the swap and immediately start the process of rebuilding the array. While there may be some slowing of the server response while the array is being rebuilt, the data is still available during the process.
Notes:
- All of the above requires planning before you even buy your server. You have to make sure that:
- You buy the right number of drives (including a replacement unit) based on the RAID you implement,
- All the drives are of the same size,
- The drives can be use for hot swapping, and
- The server and drive controller can support hot swapping.
- In some systems, it is even possible to have your replacement drive already installed in the server. It is only used when the system detects a crash of one of the other drives. The system then uses the “spare” drive to immediately and automatically rebuild the array.
