Enterprise SSD "small deals"
- Thread starter ca3y6
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Its more expensive to run older hardware for me because of power costs. 45cents/kwh
If I was being rational I would only buy just the right amount of SSD for the usage I have right now. It doesn’t make sense to pre-provision years of future usage in a market where NAND price is very deflationary over the long run. But who said I need to be rational about my hobbies?!
1. Buy the right amount of SSD, for current needs.
2. Put the pre-provisioning money into an index fund.
3. Profit and buy more SSD when you actually need it. Your money would have grown, SSD prices *may* even be lower - double profit!
Sadly, growing storage pools is a huge pain. If it wasn't for that, there would be such a big need to oversize...
Didn't raid-z expansion only make mainline in OpenZFS 2.3.0 (~10 months or so if I recall)?
Also - that means still you're inclined to grab the biggest drives you can afford whenever you create a pool. You're still driven to plan for a lot of future growth and you still pay for it up-front since the first few drives set the pool's maximum realistic size (given limited storage interfaces/slots)
I'm not sure I understand. RAID-Z expansion is about expanding the vdev, we were talking about expanding the pool itself?
To expand the pool, simply add more vdevs (of whatever kind)? Why do you need to "grab the biggest drives you can afford whenever you create a pool"?
I am actually doing the maths as I start planning my next NAS. I will use HDD as I need capacity, and I won’t buy them used (unlike SDDs, I don’t trust used HDD, at least for primary storage).
There are only so many HDD you can have in a home server if you want it to be self contained and not make more noise than a lawn mower. I think I can fit 15.
I would buy 10 upfront to form one vdev, then buy another 5 a few years later to add another vdev (as max capacities will have increased a lot if you believe the Seagate roadmap). But if you get 150MB/s effective per HDD, that means a RAIDZ1 vdev of 5 disks gives you 600MB/s, well under a 10gbe LAN speed (and I will have 40gbe). I know ZFS will also write on the original 10 disks vdev in parallel but what it won’t do is to rebalance the data between them, so logically there will be a lot more new writes going only to the newer, slower vdev, just to spread the capacity.
So zfs expansion is a bit of a pain in the arse. Since the data is backed up, an option of course is to wipe out the array and rebuild from backup when I extend it. But that’s disruptive.
Last edited:
Sandisk Raises NAND Flash Prices by 50% for November Amid Supply Shortage
NAND Flash is currently one of the most sought-after commodities in tech, alongside DRAM, due to the high demand for data storage driven by AI workloads. According to DigiTimes, Sandisk has raised its NAND Flash contract prices by 50% in November to align with the supply and demand dynamics...
raidZ speeds are significantly lower than that. Theoretical is N-1 *150MB (hdd speed) giving the same 600MB/s you wrote, but typical speeds are often half that.
As for adding vdevs to a pool, there's now (tho not feature complete til 2.4) a rewrite command that will have the effect of rebalancing a pool. I wonder if it will do any better then send|recv, but at the very least it'll be doable without an availability outage.
I can likely vouch for that. I was testing 8-wide RAID-Z2 vdevs with HGST 8TB SAS drives (~200MB/s rated), and you'd think that I'd at least get N-2*200 = ~1.2GB/s...but no. Got ~700-800MBps best case.
I've had a 4 x vdev x 6 disk rz2 pool running the same config for the last decade +. Every 24mo or so I refresh a whole vdev one disk at a time with a size bump. This has been working for me with used sas disks the whole time. (optane l2arc, 6 disk ssd 3 way mirror for meta)