Advice requested on pool topology

I am using TrueNAS as follows:
Supermicro 6048U-TR4T+
2x E5-2697v4 cpus (18c 26t ea)
128 DDR4 2400T
12x SAS3 LFF drive bays
SAS9300-8i
SAS9300-8e (connected to das)

HP StoreServ 20000 DAS
25x SAS3 SFF bays

1 pool - ZFS Z1
4x 10tb SAS 7.2k HDD
1x 960gb SAS SSD L2ARC

I have the following available for addition to that pool or a new pool
5x 960gb SAS SSD (4 more coming soon)
3x 400gb SAS SSD

I currently run on it:
Jellyfin on a Debian 12 VM (~5tb of media to serve on the zpool 1)
Frigate (container)
Home Assistant (container)

I plan to add another VM for a DB server (probably postgres)
I do not currently but am considering serving iscsi drives to VMs running on another server.

Basically I would like some advice as to what I should do with my drives and how I should configure my pool(s). I am a software dev and intend to build a massively multiplayer game that will utilize the DB and performance will be important. It could become very large over time but probably 100s of mb per day is sufficient.

I know I have some duplicated media files so it might make sense to add some dedup volumes?

I have tried to research this but the documentation I have found just says what it is and I haven’t been able to find anything really related to optimizing feature choices for target scenarios.

I would do this:

4x10TB raid-z1

5x960GB: take 4, create either RaidZ1 or z2 array. When you get the next 4, take them and add a second vdev to your 4x960GB array, use the 1 remaining as a spare
3x400GB maybe a 3 way mirror, or 2+1 R-Z1 arrray for bakcups of containers or something.

Thanks, no log or dedup drives?
Should I keep the 1 L2Arc drive on the first (4x10tb) pool?
By 2+1 Does that mean 1 spare? That is SSD, isn’t it better to use my HDs for backups?

unless you have highly duplicated data, you don’t need dedupe, also dedupe is in memory, not drives.

None of your drives are suitable for log ddevices, you would want an optane drive or something similar for that, so i didn’t recommend it. I would also pass on the l2arc drive, if you can afford more ram, do that for arc vs a dedicated drive, unless you have SOOO much data you read that you absolutely need llike 1TB of l2arc, and you’re maxed on memory.

2+1 is my way of saying a 3 drive raid-z1, since you’re loosing 1 drive worth of capacity, sorry for the confusion.

HDD’s are best for large and slow access, SSDs for small data and fast access. you could easily back up to the HDD pool, if you have anything DB related you could use the DB pool for that since it’ll have faster IO speed.

based on re-reading your post, you might want to consider a raid10 array of the SSDs if you can afford 50% loss of capacity, in favor of faster IO and rebuild on The SSD pool, and then use that for your databases, and the HDD pool for anything else.

Hope that helps!

That helps a lot! Thanks.