Why is RAIDZ1 with 6 drives (5+1) not good?

[DaveY]

Hi,

I wanted to create a RAIDZ1 with 6 drives (5+1), but everything I read says it needs to be N+parity drive where N is in multiples of 2s. I could never really find an explanation as to why though since FreeNAS itself allows you to create a 5+1 RAIDZ1 configuration with no warnings of any sort.

I have a DELL server with 12 drives and was hoping to create a pool with 2 x (5+1 RAIDZ1). Just want to know if it's safe or if I really need to have an even number of data drives. I would hate to burn 2 additional disks on parity by having to run a 2 x (4+2 RAIDZ2) configuration.

Does anyone have a good explanation as to why 5+1 RAIDZ1 is not good/safe?

Thanks

 

[dlavigne]

That's deprecated advice. From the ZFS Primer of the 9.3 docs:

Some older ZFS documentation recommends that a certain number of disks is needed for each type of RAIDZ in order to achieve optimal performance. On systems using LZ4 compression, which is the default for FreeNAS® 9.2.1 and higher, this is no longer true. See ZFS RAIDZ stripe width, or: How I Learned to Stop Worrying and Love RAIDZ for details.

 

[joeschmuck]

See ZFS RAIDZ stripe width, or: How I Learned to Stop Worrying and Love RAIDZ for details.

Nice read!

I choose RAIDZ2 for reliability which is typically the reason you see 99% of the postings stating you should use a RAIDZ2. There is some basis for it such as when a hard drive fails and especially if you are using very large hard drives (4TB or larger) and the time it takes to rebuild the new drive. RAIDZ1 is an acceptable risk to many but when some people are storing 20TB or more of data, they never want to need to reload all that data if they can help it should they have a catastrophic two drive failure at the same time (you replace one and while it's resilvering a second drive dies, yes it does happen). So having one extra drive (RAIDZ2) gives a person a little more insurance against total loss, but it is not required at all.

 

[DaveY]

Thanks dlavigne, that's good to know!

 

[Scharbag]

Yeah, RaidZ2 is the way to go IMHO. The risk associated with RaidZ1 is too high for me when using larger disks. A resilver of my system takes at least 6 hours while HAMMERING away at the vDev that is being rebuilt (and my pool is only 50% full). The increased disk loading MAY cause another drive to fail during the resilver, which in a RaidZ1 config, your entire pool would be foooked.

As drives get larger, and as you fill up the pool, resilver times will continue to get longer. Even though I back up all of my data (to a second RaidZ1 array in the same server), I would still be unhappy to have to rebuild my production pool. Just time I do not want to waste. I think it is well worth the little extra cost for 2 more drives in my setup. And this is just a home NAS. In a business environment, I would be even more paranoid and I would put my backups on a separate system in another location, not in the same server.

So, if you do not have a place to backup all of your data (never a good thing but understandable for HTPC use), I would strongly encourage the use of a 2x6 drive RaidZ2 setup for your pool.

Cheers,

 

[cyberjock]

Just to clarify Dru's post, it's *any* compression. The reason is that with compression you end up with arbitrary block sizes, so trying to align them is basically impossible.