Recommended OS SSD for 2-node S2D hyper-v cluster

[Jannis Jacobsen]

As the title says.
Any good recommendations for OS boot SSD for 2-node Storage Spaces Direct Hyper-V cluster?
Preferably enterprise ~100GB, refurbished/from datacenter is very much ok.
For lab use, so new etc is not needed.

-Jannis

 

[alex1002]

If you just using the drive for boot any good SSDs in raid 1 would work.

Sent from my LG-H831 using Tapatalk

 

[i386]

Check the "great deals" forum here, "Intel S3500 ssds" or something like this.
They are have low write endurance and performance compared to the other intel ssds but make great os boot drives.

 

[Jannis Jacobsen]

yeah, S3500 was my go to choice, as I have that in my workstation.
Just don't seem to be any good deals at the moment

-Jannis

 

[BackupProphet]

HGST 100GB SAS SSD 2.5'' HDD HUSSL4010BSS600 0B27395

 

[Connorise]

I would recommend to use SATA slots for either for cache or to increase capacity. When it comes to the boot, you could use the USB for this purpose, therefore, you could achieve more storage/cache. I'll add few links about USB boot, thus you can learn more here: Installing Hyper-V 2016 on a USB memory stick ; How-to Create Bootable Windows Server 2016 USB Thumb Drive for Installing OS | StarWind Blog

 

[Net-Runner]

To be honest, S2D completely sucks in a 2-node scenario. Even a minor issue with a single drive in each host at the same time may cause a complete data loss on such system since basically, it’s an RAID0 in each host which is being replicated over network resulting in a two-node RAID10 over the network. Simplivity https://www.simplivity.com/omnicube-product-family/ or Starwind Virtual SAN Done Right! • StarWind Virtual SAN ® are much better for such small deployments since they work on top of local RAID5/10. That grants redundancy and data locality within each single node (fault domain) in a cluster. Replication works on top meaning that if the drive fails there are enough time and resources to perform a rebalance without impacting the production. S2D usually dies almost immediately (hopefully they will fix this in RS3).

 

[Jannis Jacobsen]

The starwind vsan looks interesting.
So I could use storage spaces to create a striped pool on each server, then put starwind on top, and get "s2d" in a better working way?

 

[Net-Runner]

So I could use storage spaces to create a striped pool on each server, then put starwind on top, and get "s2d" in a better working way?

Yeah, that should work. You get a tiered or multi-resilient volume inside each host and replication is done by starwind. Looks like a great setup to me. I would still avoid using striped volumes within single node (some redundancy on a node level is just a must really). So maybe 2-way mirror or parity (unfortunately, it still sucks because of horrible performance) is better. If you have any chance to go with HW RAID - that is also an option.