1

raid-5

It's a mixture of 500GB x2 and a 320GB. (465, 465, 299.9 ish effective.)

I've never done raid before but I tried to get as close to the pool capacity as I could.

The disk space should be equivalent to 2x the smallest disk no? effectively twice 300GB..

I ended up with 810GB space and 380GB must be resiliency..

Am I doing this correctly?

Dan
  • 11
  • 1
  • I figured it out. Windows splits all storage into 3 (equal?) sized disks so data spills over. I can't be sure if I do make 300gb portions it'll be true raid-5. – Dan Feb 01 '20 at 15:35

1 Answers1

0

That is a storage spaces pool. Whether it is correct depends on your requirements.

1190 GB pool from 1230 physical is 96.7%, roughly seems reasonable assuming some metadata overhead. 810 usable in a 1190 pool is 68%, consistent for 2 parts data to 1 part parity.

Stripes are using the bigger disks past 300 GB. Would have to mean the extra space is being used in stripes with the smaller disk. See also: Mixing disks of different sizes in a Storage Spaces pool

This is not intuitive; capacities that are multiples of physical disks are easier to understand. Ideally, add same size disks in a quantity that is a multiple of the column counts.

Single parity is not ideal either. Storage Spaces Direct has the usual warning that single parity can only tolerate one failure at a time. Regular Storage Spaces apparently need at least a 5 disk dual mirror to survive two simultaneous disk failures.

So, design the overall storage system for the resilience and capacity you need. For example, a mirrored volume instead, across both 465s would be 450 GB or so usable. Not as much capacity, but easy to understand and possibly well performing.

Always back up important data. Arrays will not survive everything, nor will they prevent deleting data.

John Mahowald
  • 32,050
  • 2
  • 19
  • 34
  • Thanks for the thorough explanation. I've implemented backup procedures after deciding to use RAID-0. – Dan Feb 01 '20 at 16:24
  • RAID 0 or resiliency type simple has the most failures, because any drive failure takes out the volume. Even a single disk not in an array is better, assuming it has the capacity. – John Mahowald Feb 02 '20 at 10:36