Repository recommended max size (?)

[Cheapers]

Hi,
We have a Windows Repository server with 32 x 16 TB SATA disks which we will use configuring a small number (potentially 3) of RAID 6 sets.
Is there anything against configuring a single logical ReFS formatted 350 TB Windows partition to be used as Repository?
If so, what could be recommended in this scenario? Perhaps 2 partitions / Repositories (SOBR) to enable performance extents for faster Synthetic Full backups?
We want to avoid any complexity that doesn't provide added value.
On the other hand a volume this size just seems a bit...well, big.
Thanks!

[soncscy]

Heya Cheapers,

Depends on your needs; start with the rps.dewin.me calculator and get a "rough" estimate of your space needs for Veeam. Error on the side of caution, heavily (it's better to have more space you don't need than to end up in a position where a single error/misclick/mistake requires you delete your entire chain)

Once you have that, look to size around that; I strongly recommend find some way to build in Mirror Parity for your device (this gets into Storage Spaces =,= I already groan thinking about it), but it enhances your resilience with ReFS just that much more.

The size itself is actually fairly...normal let's say. It's definitely chunkier than the usual, but it's not in the outrageous size yet for my take.

But since you're dealing with metadata when it comes to block cloning, you want __something__ to protect you against a GP block being corrupted and then ReFS dropping an entire chain because of one bad block. With smaller volumes, you overcome this by just making redundant copies; at the 300+ TB scale, that redundant copy is a bigger ask, so finding a way to use built in technology to repair on the fly is a better investment IMO while offloading to some non-disk target.

But that's just my take.

[jamcool]

I have four setups with 35x12 TB and each in RAID 6. Not had any corrupt blocks just occasional drive failure.

Items that I would recommend are have at least one hot online spare and do RAID 60 (unless that is what meant by 3 sets of RAID 6). My biggest thing is when I loose a drive, the recovery can be very long (days or more) and performance is horrible but RAID 60 will help with this and of course how busy your system is will affect recovery as well. I also keep getting told to expect in year 3 to year 4 of using the 7,200 RPM large capacity drives to start loosing a drive a month and it will get worst. I am in year 3 of my setup and have had to 3 drives out of 140 replaced.

Sounds like got the rest of the recommendations here to with REFS and largest block size can do (64KB was biggest for me). REFS was designed to work with large disk so should be golden.