Re: Disk I/O slowness

On 8/5/26 06:24, Michael Stone wrote:
> On Tue, Aug 04, 2026 at 02:00:05PM -0400, Stefan Monnier wrote:
>> So overall the circumstances where raid 5/6 makes sense are a small
>> minority.  It's mostly for cases where total capacity is a lot more
>> important than performance and even there, the benefit is only 33% in
>> your scenario (8GB vs 6GB).  And when total capacity is a lot more
>> important than performance one would typically go for HDDs (where the
>> performance impact of raid 5/6 tend to be even more pronounced).
>
> That depends on what you mean by "performance". A large (8 or more 
> drive) RAID-6 is going to have similar performance to RAID-10 for 
> random or sequential reads (operations will be balanced across 
> drives). For sequential writes the RAID-6 will likely be faster. 
> Random writes will be worse on the RAID-6, but a write cache will hide 
> much or all of that depending on the details of cache size, write 
> size, etc. And with larger numbers of drives the difference in 
> overhead becomes more significant. (50% usable for RAID-10, 75% for 8 
> drive RAID-6, 83% for 12 drive RAID-6.) And a last really important 
> feature: RAID-6 can tolerate *two* failed drives in an array without 
> losing data. With RAID-10 if a drive fails during a rebuild, the array 
> is done. 
This is not entirely true. Suppose we have RAID-10 array which consists 
of 4 drives and thus 2 mirror groups. Array will survive the failure of 
two drives as long as they will be in the different mirror groups.
But if these two failed drives will be in the same mirror group then the 
whole array will be lost.
So RAID-10 gives 2+ drive fault tolerance as long as the failed drives 
are each in the different mirror groups. The smarter guys than me could 
calculate the probability.

> And rebuilds can take a *long* time (like, days). 
You probably thinking about RAID-0+1 array configuration, which is not 
RAID-10.
Suppose we have 8x4TB drives. In case of a single drive failure the 
RAID-0+1 array would have to rebuild the whole 16TB worth of data (one 
mirror group of 4 drives) which indeed could take a long time on a busy 
server,
but with the RAID-10 array it will rebuild one mirror group worth of 
data (4TB) which is the size of single drive.
RAID-5 and RAID-6 rebuild of 8 drive array would take a very long time, 
because not only it will have to recover (not just copy) the data from 
error correction data spread across all drives, it also will have to 
rebuild and recalculate the error correction data for the failed drive 
for the entire array.

> You can get double disk failure tolerance on RAID-10, but then you're 
> down to 33% usable space.
>
This statement looks false to me, can you elaborate?
AFAIK the RAID-10 array configuration, build from even number of devices 
(4 drives or more), will always give at least 2 drive fault tolerance, 
except in the situation I've described above, and the usable space for 
the 4 drives RAID-10 array always will be 50%.


-- 

  With kindest regards, Alexander.

  Debian - The universal operating system
  https://www.debian.org