With that size array Raid 5 is a no-no. You have a high probability of getting an URE during a rebuild, which will likely lead to you losing your array.
Raid 6 is a better bet, but even at Raid 6 you have a sufficient probability of encountering an error/second drive failure during rebuild (you can do the Markov analysis and tell you the exact theoretical probability based on rebuild times for arrays that size)
Remember, Raid is not a backup solution. Your system will likely get soft errors that will lead to data corruption over time, which is why enterprise-level servers use filesystems such as ZFS with ECC memory to check for errors.
With a raid controller you will need a battery backup to handle the write-back caching they usually do (writes will be buffered until a certain amount of data has been collected before flushing the changes to disk on the controller, if you power goes out you lose those writes)
Honestly if you are trying to build a backup server, I would use ZFS with a mirrored pool and get a motherboard with some ECC memory. This way you have protection against soft errors. You won't get the best value per gigabyte, but your data will be much safer.
For a storage/media server with a SEPERATE backup server, then the RAID6+1 approach would be acceptable...
In response to your previous question "Is there any difference between 2720, 2720S, 2720SGL ?"
After some research it looks like the 2720 standard set comes with cables, the 2720SGL does not, and I can't find much on the 2720S you mentioned.
EDIT: Of course, it really depends on how much you care about that data vs. just wanting it to be available. I handle backups for my research lab, so I am a bit anal when it comes to data integrity.
EDIT2: This will give you an idea of what kind of probabilities you are dealing with, its not a perfect calculator but its good enough: https://www.memset.com/tools/raid-calculator/