Seagate RapidRebuild claims that it can rebuild your raid upto 3x faster than the traditional RAID rebuild which sounds very promising. Before we look at how this works, lets look at 3 pre-requisits for this to work
- It will work only with Skyhawk Drives , on systems compatible with SHM ( Skyhawk Health Management). If your NVR/NAS vendor doesn't support SHM, RapidRebuild won't work and a full rebuild will instead happen.
- You must have a hot spare drive available,to which any retreivable data from failing drive will be copied.
- There should be some retreivable data on the failing drive. If the drive has failed copletely , it will be a full RAID rebuild. Time saving essetially comes from any data that can be copied from Failing drive without having to rebuild the Array.
Since RapidRebuild tries to solve the problems with the traditional rebuid , lets briefly look at the traditional raids and rebuilds
Problems of Traditional RAID Re Build
Traditionally if a drive in RAID fails, a full rebuild happens. This presents 3 large problems.
- Rebuild takes long time. Parity needs to be recalulated ( or data needs to be recalulated backwards from parity) and rewritten to the new replaced drive.
- During Rebuild the biggest risk is a secondry drive failure. In case of a RAID 5, another drive failure during rebuild means loss of data.
- Performance is degraded during rebuild.
On a sidenote, RAID 5 + Hot Spare is always a bad idea and its better to go with RAID 6 instead.
Risk of the Secondary drive failure is the biggest risk as it can make you lose your data if you have single parity RAID like RAID 5. The Sooner the RAID rebuilt is done, the better. This is the time that Seagate RapidRebuild it trying to cut down.
Minimizing RAID rebuild time with Seagate RapidRebuild
The way RapidRebuild can reduce the rebuild time is by relying on its capability to predict a drive failure and copy as much data as it can on a spare drive. Since copying is a much faster operation compared to rebuild ( which involves calculating the parity again and writing to the disk), it cuts down rebuild time.
Consider this -SHM ( Skyhawk health management) system knows when a drive is failing in advance. Since it knows the drive is soon going to fail, it still has good chance of retreiving a good amount of data from the drive, before it actually fails. Here is what happens with RapidRebuild
- SHM predicts a drive is going to fail.
- SHM starts to read whatever data it can from the failing drive and starts to write it to the hot spare. Since copying doesn't need parity calculation, its very fast.
- lets say 80% of the Data is copied from the failing drive to the new drive. Now the rest of the 20% data will have to be rebuid.
- 20% of the data that couldn't be copied is rebuilt.
- You have a RAID with the new drive , the old drive is marked bad and can be removed and replaced with a hotspare.
How much time you save with RapidRebuilt depends on how much data could be retreived from the failing drive. If you had a drive that completely fails, all the data would have to be rebuilt with no time saving.
ADAPT by Seagate ( Advanced Distributed Autonomic Protection Technology)
Since we are on the topic of RAID, its worth mentioning that seagate also has ADAPT technology which is used in Exos Systems. ADAPT technology stripes data across multiple drives, adjusting automatically and keeping hot spares all the time. Advantage of ADAP is that its self healing and builds RAID very rapidly.
In the Disk Arrays, RAID is essential and everyone is trying to find ways to make it more efficient. With Erasure codes, propritory RAIDs and alternative to traditional parity bits, there are many ways to enhance the RAID.