Objective hard drive configuration for office server and file system organization?

Question

Server optimization

There is a freshly assembled server. Assembled independently, according to your needs.
The configuration is this:

• CPU: 2 x Xeon X5667 socket 1366
  
• MB: supermicro X8DTL-i
  
• RAM: 2 x 8Gb ECC
  
• HDD: SATA-3 1Tb 7200rmp (regular desktop)
  

The question of old hardware ... With the introduction of DDR4 - tomorrow everything modern today will become not modern, because everything is so sad, and an upgrade is not expected, I don’t see the point in new and expensive "modern" equipment, it's cheaper to take the old one, fully equip it and live next 5 years quietly.
It costs CentOS 6.4 (temporarily on a SATA disk)
The source of the issue is that it was not possible to get the desired mother, I installed what I had to, overlooked the absence of a SAS controller on the motherboard.
The question of how to optimally distribute the "system and content" among the disks was especially acute when there was one SAS disk 300Gb, 15000rmp, which, in the absence of a controller on the motherboard, is dead weight on the shelf. It is possible to return/exchange.
Planned:

• Mail server
  
• Corporate website with a database for employees + Online store (in the future) (!)
  
• Employee file space available over the network
  

The question is which configuration do you prefer?

• Option number 1: 1xSAS for the system, 1xSAS for network services (for an online store, etc.), 1xSATA for user files, 1xRAID controller for connecting SAS drives in JBOD mode. The system in the image, the rest in the backup in case of disk failure. Issue price: + 15t.r. to the cost of the server.
  
• Option #2: 2xSAS in RAID 1, 1xSATA, 1xRAID controller. System on SAS, user files on SATA. As a result, increased reliability, but a greater load on disks due to network services. The image of the system is not needed, everything else is the same as in option No. 1. Issue price: +15t.r.
  
• Option #3: This is the same option #2, but adding an SSD for network services (you'll have to think about email files and attachments, since they definitely don't belong on an SSD). As a result, we have the possibility of depriving daily information due to the failure of the SSD. As an option, frequent backup from an SSD drive. Issue price: + 17t.r.
  
• Option #4: This is the same option #3, but plus one more SSD in RAID 1 to the first drive. The chance of losing information when an SSD fails is reduced. Issue price: +19t.r. I wonder how many SSDs will die in a year and is it worth bothering with them at all?
  
• Option number 5: The most incredible, because the level of greed on it rolls over, but probably the most optimal. 4xSAS in RAID 10, 1xRAID controller. User files, as always, are on SATA, everything else is on a disk array. Issue price + 27t.r.
  
• Option #6: PC Level Option. 2xSSD in RAID 1 for the system, but mount /tmp /var /swap on 1xSATA (a small amount of 300-500 GB, although the price tag of 1TB is not that far), /usr /home mount on, 1xSATA 1TB and there will be files on it users.
  

Answer 1

It depends on the case and the current loads from your list, I would do it on option 2, only network services on SAS, user trash on sata, if there is not enough performance, you can always plug in an additional SSD as a cache.
No more than SAS disks within the margin of error, now SSD disks are already quite reliable, well, backups have not been canceled, if anything, a raid is not a replacement for a backup in any case.