Performace hit - 5HDD's v.s. 6 HDD's - RAIDZ2

[TDPsGM]

I have been doing a bunch of reading - just finished going through the 186 pg manual for the 8.3.0 Release - plus looking over the Forum and making notes.

I understand that it is 'optimal' to do a 4,6, or 10 HDD Raidz2 configuration, but there is nothing stopping one from doing a 5 HDD Raidz2 config.

IF I did a 5 HDD Raidz2 configuration, what exactly am I looking at as far a a "Performance" hit?

Is it a READ / WRITE speed issue and if so just how much of a speed hit am I going to take (if there is a simple answer to that, that is)?

ALSO, (I don't remember where it was I read it in the forum) but it sounded like Raidz2 wasn't considered efficient until you got equal to, or over, the 6 HDD configuration. If I read that right, what is the reason for that again?

Thanks

 

[TDPsGM]

Section 6.3.3 Volume Manager of the 8.3.0 Release Manual doesn't exactly discourage it:

if you select five disks, you can create a UFS or ZFS stripe, a UFS RAID3, or a ZFS mirror, RAIDZ1, RAIDZ2, or RAIDZ3

 

[William Grzybowski]

The result will be not aligned read/writes with blocks not equally distributed between the disks, one of the disks is going to hold extra data probably lowering the IOPS a bit.

The hit might not be substantial depending on your load, you probably wont even notice anything if you're network bound at 1Gbps.