dedicated root aggregate

[superderivatives]

Hello

I really don’t understand why we need to have dedicated disks, Raid and aggregate for the root-data in FAS8xxx clustered mode.

Dedicated raid root aggregates for each node of a Clustered Data ONTAP cluster is a huge waste of disk space, spindles, shelf space and our money. And for what? For containing the log files. Even not the OS core files.

This is unreasonable and annoying.

I will love to read your comments on this subject, as for now, I am not going to pay it.

[ScottMiller]

I'm with you brother.  I have root taking up 18TB of disks (3TBx3x2) on one of my pairs.  That's a lot of capacity I could use elsewhere. 

The advanced partitioning in 8.3 solves this problem, but apparently only in 2500 series and all-flash configurations.  I don't know why this isn't the solution for all models, but maybe if we make enough noise....

[cgeck0000]

I remember back in the day reading that if there's some type of corruption in the aggregate and the system has to do a chkdsk it would take the controller down, because the root is on the same aggregate as the data. So separating the root vol out to its own aggregate would minimize that.

I personally have never placed the root volume on its own aggregate and haven't had an issue in 5 years of working with NetApp storage. Haven't heard of any customers I've spoken with have an issue.

[superderivatives]

Still doesn’t make sense.

1. The OS is on the flash card. Only the logs and maybe configuration files are on the root vol. why not extending the flash card with +50GB?
2. They do have a partial solution with the 2xxx series: partitioning. Why not do it for the 8xxx too?

In my case I am wasting 3x900GB disks space per controller. (imagine you have 8 filers in a cluster. huge waste of space, spindles and money)