4-bay NAS write performance RAID 5 + SSD vs RAID 10

Question

I am planning on buying a new NAS (QNAP TS-453 Pro) for my Home Office. The NAS has 4 HDD bays and supports SSD Caching. Hence, I was wondering which setup would probably be faster in terms of write performance:

3x4TB in RAID 5 + 1 SSD (512GB - 1TB) as cache or 4x4TB in RAID 10

Typical tasks would be backups on iSCSI targets and file transfers to network shares. The max number of clients performing write operations will be 3-4.

Both proposed setups would have 8TB of available storage, with the RAID5 + SSD solution being a bit more expensive.

Is it possible to say which setup will be theoretically faster or is trying it the only option to determine the difference?

Answer 1

With 4 TB disks you probably shouldn't be running RAID5 as the odds that you'll be running into an unrecoverable error when rebuilding the array are significantly increased. From a reliability perspective I would recommend the 4 disk RAID 10 configuration.

Using the SSD as a cache will of-course be faster in theory and probably in practice as well, at least for some workloads, but at the risk of losing the remaining array during the very IO intensive RAID5 rebuild after a disk failure when you have replaced a faulty spinning disk.

When your primary usage is to hosts backups on the NAS the main IO characteristic will be sequential writes for which spinning disks are often already fast enough to that a common 1 Gigabit/s network link is a bottleneck before the speed of the disks are.

If you need both performance as well as capacity, why not create mirror of 2 large capacity disks for the first volume and use that as the "lower performance" file share, and create a second mirror of 2 SSD's for tasks that really require a high number of random IOPS (i.e. a iSCSI target that hosts a number of disk images for virtual servers).