What's happening with disk rebuild times? How do 'preemptive rebuilds' work, and what other techniques are appearing, if any, to accelerate rebuild times? |
 |
| 16 Nov 2007 | Greg Schulz |
|
|
On one front, rebuild times are actually getting better. Now, before I go off on the limb on that one, as drives become faster and rebuilds on the controllers become faster, we're getting hit by growing capacity. So, the net result is that any drive rebuild improvement is being offset by this capacity growth.
So, in some cases, products use preemptive rebuilds to eliminate known bottlenecks in the rebuild by proactively performing a rebuild or copy operation. The risk is false positives. Incorrect diagnosis could cause excessive rebuilds or copy operations. So, while preemptive rebuilds could be a good thing, you need to do your homework to find out if the preemptive approach is really a workaround for some other known limitation.
Ask the vendors how many concurrent rebuild operations can take place. Also, ask about the impact of those rebuild operations on the normal running workload.
Check out the entire RAID FAQ guide.
|
|
|
|
