Does a slow parity hdd slows the faster nvme disk in use?

[HackHome]

I have the impression that the slowest disk (my hdd) used as a parity disk is slowing (indirectly) the use of my nvme which I use to run VMs on. Is that true? Or just my impression?

Like installing windows 10 took very long now and on the disk main page I could see write speeds as low as 11-12MB/s instead of the speeds that the nvme should be capable of. Is it like a mirrored raid that all data written on the nvme also will be written to parity disk and as this disk is the slow one it automatically slows down operation of my VM running from nvme?

[JorgeB]

If the NVMe device is assigned to the array it's true, any writes will be much slower.

[HackHome]

Well how else could I use the NVMe device without assigning it?
The hdd is the one and only parity disk.
The NVMe is the one and only disk in the array storing my data.
What do I do wrong or is there anything else I can do to accelerate the use? Do I have to use an nvme SSD for parity as well to eliminate the bottleneck?

Or to sum it up: Will the array never be faster than the slowest part of the array (being it an active used disk to store data on or used as a parity disk)?

Thanks so far

Edited September 20, 2022 by HackHome
added question

[itimpi]

31 minutes ago, HackHome said:

Or to sum it up: Will the array never be faster than the slowest part of the array (being it an active used disk to store data on or used as a parity disk)?

This is true as far as writing to the array is concerned.

 

if you have SSDs then these are normally used in a ‘pool’ external to the main array.   Pools can be made redundant by having multiple drives in the pool, but you are still constrained by the slowest component.   There is little point in having a SSD as an array drive with a HDD as parity, particularly with the way Unraid updates parity as described here in the online documentation accessible via the ‘Manual’ link at the bottom of the GUI or the DOCS link at the top of each forum page.

 

With the drives you mention you could have a single drive array using the HDD.    Then you could have a single drive pool using the SSD to host VMs.docker containers and run them with the full performance offered by the SSD.   Since this configuration does not provide redundancy then you could could do periodic backs (frequency chosen by you) to make backups of the VM/docker files from the SSD to the HDD drive in the array.   Plugins are available that can automate this process.

[HackHome]

Thank you a lot for the explanation, details and hints. I will work my way through docs.
So as I understood pools CAN be configured to behave like RAID 1. The advantages of using a pool in unraid is that I can add (fast) disks anytime to the pool so it increases flawlessly?

Sorry for all these newbish questions.

[itimpi]

Pools can use any of the pseudo RAID levels supported by BTRFS.   It is a BTRFS specific implementation of RAID that can be dynamically expanded, can change RAID levels and can use odd numbers of drives.   The downside is that BTRFS seems to be more susceptible to file system level corruption than XFS (which is the default for the main array).