General storage (with redundency).

(Specs in sig)

Just to preface, any important data will also be copied to an external drive for additional protection.

I'm thinking bout setting up some general storage in my rig with some sort of redundency.

Am i better off with just 2 drives in Raid 1 (windows or FakeRAID?)

Or 3 drives in a Raid 5, again, windows Raid (soft) or FakeRAID?

Performance is secondary, my 1TB drive is for games (steam).

Raid 5 will give you more capacity.

If I have to pick between RAID 1 and RAID 5, and capacity isn't a factor, it's RAID 1. Why? Simply because failed drive recovery can be as easy as just removing the drive from the raid and accessing the data.

RAID 5 however requires a rebuild before you can see the data again. And if you have a second drive fail during that rebuild--poof! All data gone.

Bit of a reply to a necropost, but this isn't quite accurate:

SamirD said:

RAID 5 however requires a rebuild before you can see the data again

Click to expand...

RAID 5 arrays with a failed drive are still accessible, although typically at a degraded performance level (depends on the controller really).

Your decision has to be how important the data is and how many disks you're willing to give up for redundancy. For my 'big storage' I run 8x 8TB drives in RAID6, which gives 2-disks worth of failure tolerance. I do this because I have a lot of data - media mostly - and losing it all would be a huge pain in the ass. It's not a big enough pain in the ass that I back it up - mostly because I've got no way to effectively back up 15+ TB of data offsite - so I focus on internal redundancy to the best of my ability.

You also have the option of something like StableBit DrivePool, which can utilize multiple drives and specify how many redundant copies of your data you want, without having to use a RAID controller.

sinisterDei said:

Bit of a reply to a necropost, but this isn't quite accurate:


RAID 5 arrays with a failed drive are still accessible, although typically at a degraded performance level (depends on the controller really).

Your decision has to be how important the data is and how many disks you're willing to give up for redundancy. For my 'big storage' I run 8x 8TB drives in RAID6, which gives 2-disks worth of failure tolerance. I do this because I have a lot of data - media mostly - and losing it all would be a huge pain in the ass. It's not a big enough pain in the ass that I back it up - mostly because I've got no way to effectively back up 15+ TB of data offsite - so I focus on internal redundancy to the best of my ability.

You also have the option of something like StableBit DrivePool, which can utilize multiple drives and specify how many redundant copies of your data you want, without having to use a RAID controller.

Click to expand...

Ah sorry! You're right--that was completely wrong. The data in a single failed drive RAID5 is still accessible. I don't know what I was thinking.

This is the second time I've seen stablebit drivepool recommended as a storage solution. Have you tried it, and would you mind explaining the gist of it so I don't have to wade through all the marketing material?

SamirD said:

This is the second time I've seen stablebit drivepool recommended as a storage solution

Click to expand...

If you're talking about on these forums, I'm afraid it's entirely possible that both times it was mentioned were by me, I seem to recall mentioning it a few times recently.

I've not used it beyond the trial, as I decided to stick with my traditional RAID 6 since I walked into the drives and controller. In general terms though, it works by creating a virtual hard drive, and when you put stuff on that virtual hard drive it magically under-the-covers actually distributes those files to X number of real hard drives you have joined to the pool of drives. There are a few magic bits here. Firstly, files aren't broken into bits and striped across drives like they are with RAID - they're stored as complete files in regular NTFS folders. That means that you can plug each of the "real" drives into some other system one at a time and copy all your stuff off of them, without having to have access to a copy of DrivePool or some proprietary RAID controller to read the data. Secondly, you can set your own level of redundancy - maybe the virtual drive you're storing your most precious family photos on you say "I want 3 levels of redundancy" but for your porn collection you say "1 level of redundancy is fine". This can also be adjusted on the fly. For RAID, you obviously don't get to set this at differing levels for differing data, you're stuck with whatever RAID level you built the array at. Thirdly, you can use different sized drives, as well as add and remove drives from the pool without having to rebuild the whole thing. More drives mean more space, and you can mark a drive to be removed from the pool and it'll evacuate data from it and move it over to other disks.

In terms of downsides, there are a few. It gets none of the speed benefits of RAID or RAID controllers, which may or may not matter to the individual. Additionally, and this just bothers me, each individual drive still shows up in Windows as a drive letter, and then the virtual drive shows up as a letter as well. In other words, for MY deployment I'd have eight damn disks in Explorer plus a ninth for the virtual disk, which is just bonkers.

Thank you for the explanation. Sounds very similar to softwares of yesteryear that tried to do the same thing. I've actually stopped using RAID entirely and just mirror the data manually myself as well as compare it regularly for bit rot--that's what I've found to be the really silent killer of data in the last 10 years. Having a minimum of 3 sets of data can help you weed out the 'bad' file and put a good one back in its place, but it's a bit tedious to say the least.