-2

I have purchased 4 drives from Seagate to build a home cloud based on RAID5. However, I occasionally ordered different models: two ST4000VN008 and two ST4000VN006.

Now I am wondering, shall I send 8's back and order the newer model ST4000VN006 with the larger cache instead? Alternatively, maybe model with the higher spin rate (ST4000VN008) would be better for my purpose?

I would opt for less vibration and heat.

Finally, maybe it does not matter and these two models will work together just fine?

Dmytro Prylipko
  • 97
  • 1

1 Answers1

1

Don't do this please - RAID 5 is essentially a dead technology and storage pros have been avoiding it for well over a decade now - especially on large SATA disks - you will lose data and we have users come here very frequently asking how to recover their data from arrays just like this.

Use R6/60 or R1/10, or RAID-Z if you like it, but please don't do this with R5 or we will see you back here on a less happy day.

Either way you've be ok to mix disks - oh and we don't answer home questions here anyway but I needed to let you know how dangerous R5 is - many here think it's borderline negligent for vendors to still offer it.

Chopper3
  • 100,240
  • 9
  • 106
  • 238
  • Thanks, I am reading the article about RAID5 right now. I have choice between 0/1/3/5/10. RAID10 looks the best option, if I want to secure my data somehow. Also, I can bundle two 8's into set A and the two 6's into set B. – Dmytro Prylipko Aug 26 '22 at 11:33
  • BTW RAID-Z is practically an analogue of RAID 5. Maybe you mean RAID-Z2 (or Z3)? – Romeo Ninov Aug 26 '22 at 12:03
  • 1
    @RomeoNinov - you're probably right, I'm no expert on that sorry – Chopper3 Aug 26 '22 at 12:31
  • 1
    Single disk redundancy of RAID-Z is still not great given the chances of error during rebuild. However ZFS (and Linux MD and hardware RAID) have various ways to mitigate other problems like write hole, each with their own pros and cons. Server Fault has discussed these before: https://serverfault.com/questions/844791/write-hole-which-raid-levels-are-affected – John Mahowald Aug 28 '22 at 17:45