RAID 60 for the win Very large disks can have very long rebuild times so we risk another disk failing during rebuild - to mitigate this we pick RAID 6. RAID 6 comes with a huge performance penalty so to mitigate this we use RAID 0. In combination, we end up with RAID 60.

I like this, and it does work well with scales that justify the density, follow up question for your opinion, would you sacrifice some capacity to specify hot-spare disks as well? This always splits opinions when I ask the question to a customer

I also go with RAID6/RAID60; especially for backup storage those offer the best price/capacity ratio, while still being more resilient to disk errors. RAID6 vs RAID60 depends on the disk size and disk count (not discount^^).

@MicoolPaul I rarely see hot-spare disks anymore. With a decent support contract, the replacement disk will be delivered very quick and together with dual-parity, a failed disk isn’t so critical anymore. I rather use the disk bay for capacity

Our storage arrays we use are typically set up with RAID6 for our disk pools.  We mainly use Hitachi and Lenovo at the office.  At home I have Synology DS920+ and that is RAID5.

We use RAID6 in most cases.

On some smaller disk repos based on internal server disks RAID5, because there are not that much disks.

We don't manage big Luns or Datastores, and for us protection / performance works great with Raid 5, with a Hot Spare Disk when possible / critical Storage.

RAID10 for us as we don’t utilise a lot of storage and it gives us just the right performance for our workloads.

@HunterLF and @dips , you say that you don’t have a lot of storage. Do you put all or most of your backup data in the cloud?

Just being curious…

Not Really buddy, we handle 4 TB more or less, so in terms of Storage, We use for backups something around 8TB to 10TB.

Im doing a POC with Wasabi, amazing…. it's all I can say, 
and now struggling with financial to get the budget approval to keep it!

Also working in a deduplicated repository to store the Archiving backups, the GFS with un to 5 years retention in a 5TB Virtual Appliance, with the feature of 20:1 dedup ratio (dream… would be awesome to reach 10:1).

Ask and be curious any time you want! also you can always reach me to PM, 

feel free to contact me, any of you, 

would be great also

Friday or Saturday night virtual Beer!!

@Madi.Cristil please!!

Im going right now to my favorite Pub to have one!

cheers!

Great weekend!

Poll results looking promising so far! No one is storing their backups on RAID 0

Well with 3-2-1 it shouldn't be a problem to store backups on RAID0...perhaps we're doing it all wrong?

And, no one said it, but RAID isn't backup.

THERE WE GO! Wanted someone to say that

No, but a good base for your backup data

Hey @HunterLF ! You could probably organize something like that, I bet people would join a Friday night virtual beer '’event"! Count me in :D

RAID6 is used in most cases. 

Raid 60 ftw like @haslund , 128kb strip size and 90% write 10% read for raid controller cache

We tend to keep as many restore points as we can on premise and the rest is offloaded into the cloud

OK, thank you. I was just curious, because you said that you don’t manage much storage.

For my customers the storage gets more and more...

I think we are planning on adding more onsite capacity to keep more restore points and to speed up recovery times. 

As the years go by we accumulate more and more data. 

Hi,

It all depends on backup data type, usually we use RAID 6 which provide more resiliency but if you go for other RAID options like 6+0 etc then you have to sacrifice your storage capacity so it’s all about how much storage capacity you have and what kind of backup you want to do.

Another important metric for choosing what type of RAID configuration to go with is rebuild times if a drive were to fail. With drives getting larger and larger, it can easily take more than a few days after a disk replacement to rebuild the array, which in turn puts pressure on the other drives. 

As mentioned above in the thread, RAID is not backup so have at least one backup copy offsite as a minimum, on a different storage medium. 

RAID10 on the primary VBR server.

Then having two offsite BCJ servers on each their own location also rocking RAID10.

Combined storage RAW is 504TB, after RAID10, bit/bytes conversion its 234TB usable in Windows.

You could argue the 14TB disks are expensive, but then again - backup is insurance of company data that thousands of employees has spent decades creating - what is the cost of that?

And if we cant restore the day it matters the company is gonna die.

RAID10 all day.

Also the security around the servers must be high in regards to access, OS lockdown, not on the company domain, certificates, encryption, firewall, MFA login etc. etc.
 

There are many articles, discussion etc. online regarding large (4TB+) disks, URE, parity rebuild times on RAID6/60 being weeks on 8-12TB+ disks. I have experience a RAID50 not being able to rebuild a broken disk. Once you realize the array is not coming up...

The three servers we bought is apx. (investment with carepack with a lifespan of 5 years) apx. 700 USD per month (plus electricity I know, but comon) - that is pretty cheap considering we are safeguarding 234TB of data, on prem (two offsite, but its our offices), securely and the backup and restore performance is 1:1, not throttled etc.

I know not everybody has the budget etc. but at least when entering the larger disk league please rethink the RAID6/60 scenario - rebuild times, URE etc. could be a dealbreaker for you.

We tend to use RAID6 on our Hitachi arrays when the storage gets set up.  Standard that they have used for years.  But also, data dependent as most have said above and depends on the use case too.

Flash SANS with DRAID. insanely fast rebuild times, which is good for 40TB flash core modules.