Hey guys,

 

Quick question.

 

Whats better? Single 240gig SSD in cache or dual 120gig SSD in cache?

 

Just wondered if theres any performance increase to be had from having another hard drive to the cache pool or not.

 

Thanks for any info put forwards.

 

Cheers

SSD is not a hard drive.

 

From my experience, BTRFS RAID-0 carries no perceivable real life benefits. So given the risks of things going wrong with RAID-0, single is better.

 

And that's before we consider the on-going issue with btrfs multi-drive cache pool performance.

Thanks testdasi.

 

Ive just clean swapped out my motherboard on my unraid server to a core i3 6th gen board with NVMe capability.

 

Would i be better off using an NVMe stick in cache or storage?

 

Thanks 🙂

Edited April 22, 20206 yr by Andrew Clarke

Anyone had any experience with running NVMe drives in an unRaid configuration?

 

Just looking to find the most optimal performance benefits from running NVMe drives in an unRaid setup.

 

Thanks guys

Sure. Running two M.2 Samsung 970 EVO building a BTRFS Cache Pool. All Dockers and VMs are running on this pool since over a year or so.

 

***Edit*** I'm using two Lycom DT-120 PCIe x4 M.2 Adapter to connect these two M.2 drives to the Motherboard.

 

Edited April 23, 20206 yr by hawihoney

20 hours ago, Andrew Clarke said:

Would i be better off using an NVMe stick in cache or storage?

What exactly do you mean by the word "storage" here? You don't want to use SSDs in the parity array.

13 minutes ago, hawihoney said:

Sure. Running two M.2 Samsung 970 EVO building a BTRFS Cache Pool. All Dockers and VMs are running on this pool since over a year or so.

 

Would I see any measurable peformance increase over using NVMe M.2 drives in comparison to running say SATA M.2 drives guys?

Don't bother with SATA M.2. It's just a waste of SATA port with M.2 form factor heat issue.

 

You are asking short generic question without specific details which I suspect won't yield any useful answer for you.

 

Also "measurable" is entirely different from "perceivable" and "applicable".

It can be measurably faster but not perceivable in your case because your use cases are not applicable.

1 hour ago, Andrew Clarke said:

Would I see any measurable peformance increase over using NVMe M.2 drives in comparison to running say SATA M.2 drives guys?

I did start with two SATA SSDs building the cache pool. Changing the cache pool to two M.2 NVMe drives connected to PCIe x4 was a HUGE difference. I'm no hardware guy but if I remember correctly the difference e.g. in read performance was something around 550 MB/s vs. 2700 MB/s. 

 

Thats what I thought hawihoney thankyou. 

 

Appreciate that the throughput is a questionable factor but thats basically the information I was after. 

 

Yea its a bit of a no brainer when you think about it. I meam 500MBps vs 3GBps is a substantial increase in performance. Now I know what Im looking for NVMe drive wise, I shall be looking to buy a minature 10gigabit switch and two 512gb NVMe drives and then look to rearrange my whole setup so its a bit cleaner and tidyer. 

 

I imagine that I would want to run the NVMe drives in the cache pool and use SSDs for storage?

 

Thanks for your help guys

IMHO, I would not put SSDs in the array. Put VMs and Dockers on the cache pool and use SSDs or better NVMs for the cache pool. Use the array as storage. And use spinners in the array.

 

Just my 0.02 USD.

 

 

 

 

A welcome $0.02 thank you HawiHoney

 

I shall endevour to do this when I get some time free.

 

I've just bought a spare z270m motherboard and plugged a spare i3-6100 sat round doing nothing in to it for better IOPS (was running i7-3770k), but that was on a DDR3 Z75 Pro3 motherboard.

 

Thanks for the feedback. I'll be pinning this discussion and using it when I purchase the 10gig switch and dual NVMe drives.

 

Thanks