Best use of 4 disks in RAID


Recommended Posts

What is the best use of 4 disks which gives me most storage but with some form of data redundancy? I was just going to mirror 2 disks to the other 2 but that's obviously not the best use of space. Can striped with parity work with 4 disks? I was always always under the impression its a 3 drive operation (stripes on 2, parity on 1) but would like to be corrected.

Thanks

Link to comment
Share on other sites

RAID 5 can work with 4 disks so that it (stripes on 3, parity on 1) but the stripes & parity are shifted around across 4 disks but still allows for 1 out of 4 disks to fail with the data protected and 75% of the total space out of 4 disks.

Link to comment
Share on other sites

It really depends on what you need out of it. If data integrity is the absolute most important then RAID 1.

If you need integrity and more space than a single drive offers, RAID10. RAID10 will give you the highest level of data redundancy (a complete mirror!) but not as much space as RAID-5. But think about it... one drive fails, and you're without your data for hours to days while it rebuilds (if it completes...)

I don't recommend RAID-5. It's becoming more likely to fail. Raid-6 if you MUST, but don't RAID-5.

http://www.smbitjour...-more-reliable/

http://www.smbitjour...-or-a-hot-mess/

RAID5.jpeg

Link to comment
Share on other sites

RAID5. But if you later expand the array to 8 disks or more I'd highly recommend changing it to RAID6. And always backup, raid isn't a backup solution do not rely on it to safeguard your data, only to cut down on down time in the event of a disk failure.

Link to comment
Share on other sites

RAID-5 is probably the best option. RAID-0+1 would just half the amount of storage you have available. As for the comment above about RAID-5 reliability, software RAID is risky business anyways.

Link to comment
Share on other sites

RAID-5 is probably the best option. RAID-0+1 would just half the amount of storage you have available. As for the comment above about RAID-5 reliability, software RAID is risky business anyways.

The first link was for you then ;)

6TB RAID-5, 50% chance of URE. 12TB, 100% chance. 3TB array, 25%.

Link to comment
Share on other sites

"6TB RAID-5, 50% chance of URE. 12TB, 100% chance. 3TB array, 25%."

I don't think your doing the math right..

http://www.raidtips.com/raid5-ure.aspx

This scare mongering has been around for many many years.. Raid 5 was suppose to be dead in 2009, its mid 2012.. So something is not quite right.

I agree with you that there are lots of articles saying that raid 5 is dead.. But strangely its not ;)

What I would suggest, is if you have data that your worried about it being offline - which is the whole reason of raid. That you feel you should run raid, I would make SURE you have BACKUP!!! Raid is not a BACKUP!!

Raid with parity is a method of mitigating having to restore from backup, on minor hardware failure (1 disk out of many).

What disks do you have? Not all disks are good choices for raids, and yes a rebuild can put a strain on your other disks as they are hit to rebuild the failed one. So if they are not in good shape, its possible they could fail on a rebuild.

As mentioned in the article I linked too that points out what is being said might be a bit exaggerated lets say. Lets look at this statement from your link

"the rate of URE is 10^14, or once every twelve terabytes of read operations. That means that a six terabyte array being resilvered has a roughly fifty percent chance of hitting a URE and failing."

I don't buy that, for starters sure a URE could be a major problem - but I don't buy his math on the likelyhood of hitting one. For starters who says there is 6TB of data on my 6TB array that needs to be rebuilt? Do you run your ARRAY at 100% Cap?? So your 6TB array has what being used 3TB? Which is stripped over all the disks in the array right. So for 6TB of data your talking 4x2TB (you loose 1 disk for parity in raid 5 right - which is striped across all disks)

So my 3TB of data is striped across 4 disks with the added parity your talking 4TB of written stuff (3TB of data and 1TB of party), or 1TB on each disk. So to restore that data you have to read 1TB off each disk that is left in your 4disk array. The raid magic then happens and using parity recreates the 1TB that is gone off the failed disk.

So if each disk could have a bad sector that could cause a URE - what are the odds that its in the 1TB I am going to read off each disk? I don't think its as simple as they are saying that a URE happens in every 12TB of read data.

What I am trying to say is don't take these Doom Sayers as gospel.. If that was true - raid 5 would of been dead back in 2009 like they were saying ;) Its 2012 and raid 5 is still here. Like they are saying raid 6 will be dead in 2019 ;)

edit: As to the OP question at hand - even though I think raid 5 is still a viable option. I would prob not go that route myself, there are many many different options these days with newer ways to use your diskspace while still providing parity and ability to rebuild data on loss of disk. Unraid comes to mind, snapraid, flexraid. There are also methods of just keeping your critical data on your storage on more than one disk, while not having to create parity for every single bit. Drive extender, DrivePool from stablebit, etc. etc. Im currently using drivepool for my storage needs. It allows me to assign folders in my storage that need to be stored on more than 1 disk, so that if 1 disk fails I don't have to worry about restoring those files from backup.

Because most of my data I would never backup or even worry about restore if lost, but some of it I do backup!! And is critical - so I just keep multiple copies of it on multiple disks in my storage array.

Link to comment
Share on other sites

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.