Dell R730xd configuration check

First of all, thanks to the community here! I’m quite new to truenas and read about some first configuration steps a lot. This has already taken me a good step forward in terms of planning my server with TrueNas. This has already allowed me to gain some initial experience and carry out a few tests.

Nevertheless, I would like to get a few opinions from you on the pool/vdev layout as far as my exact configuration is concerned. I am also referring specifically to the configuration of raidz or draid. Also whether it makes sense to use the optional log, spare, cache, metadata and dedup disks.

Here is how my server looks like:

  • Dell R730xd
  • 24x bays 2,5”
  • 2 rear flexy bay drives (disk 25 and 26)
  • Perc HBA 330 mini controller (the enclosure works in unified mode)
  • PCIe SSD Controller Card (which can be used for U.2 NVMe SSD’s in slot 20-24)
  • 12 x 600GB 10K HDD
  • 2 x 300GB 10K HDD (in the back)
  • 64 GB RAM
  • 2 x E5-2620 v3
  • Intel X520 10G 2P

TrueNas Scale 24.10.2 is already installed and basic configuration is done.
I want to use the server with TrueNAS installed for a lab with 5 Dell R720 hosts and about 70 virtual machines. There is vSphere installed and the connection should be done via nfs share.
About 60 VM’s are running but I think only 20 are really used every day. The rest is just bobbing along.

I really like the option of being able to install U.2 NMVE there. But that won’t be the case for another 2 or 3 months.
I currently have the BootPool installed on the two 300GB at the back.
The remaining 12 x 600 GB HDDs are running a simple RaidZ3 for the time being, without separate disks for log, spare etc.
I have already carried out initial tests with it and am actually satisfied with it. However, I now have the option of replacing the 12 HDDs with 12 x Samsung SSD 870 Evo 1TB. We still have them and I’m wondering whether I should do this, as they are actually consumer SSDs?
They would definitely be faster and certainly offer noticeably better performance, especially for the VMs. So far we have the VMs running on 3 different data storage devices (all HDD only) and currently have a total disk usage of just under 5 TB. So rather manageable.
So depending on the Raid configuration with the 12 Samsung SSDs, there would be enough space. There would still be 8 free slots, which I could equip with normal hdd’s again.
And then there is the option of adding U.2 SSDs in 2 or 3 months’ time.
Has anyone with identical or similar hardware ever implemented this before?

I have the following questions about the configuration:

  • Is an HBA 330 sufficient for this in terms of speed?
  • Is a 10GB SFP+ connection sufficient? At the moment it is no more possible.
  • should I change the two 300GB HDD to a consumer 120GB SSD from Samsung?

Does anyone have a suggestion as to how I should best allocate the disks so that there is also some performance for the virtual machines? Possibly like this?

SSD pool with 5 x Mirror VDEV’s, so that a striped mirror is created and 10 disks are used. In total then 4.4TB of data. One SSD still as spare.

HDD pool with 3 x mirror VDEVs, resulting in a striped mirror with 6 HDDs (= approx. 2.2TB). 2 HDDs as spare and the last SSD as log.

Or would you prefer 4 x mirror VDEVs for the HDD pool? I can quickly replace a defective disk if necessary. I also already have an external backup. It is also sufficient in terms of storage space. And I may add another 4 x U.2 NVMe at some point. But the price is not easy right now.