On 20 Jan 2009, at 19:08, Milazzo Giacomo wrote:
Hi all
Ok. Before to have a mess of different opinions I want to definitively explain again because I wrote “nonsense”, with a smile, remember it J Very ironically…
Most of us well know what aggregates, flexvols, raid dp or raid 4 are and we don’t need to explain. Most of us know that spares, parity disks are not “waste” space for they save your data and very oftern they save more! ;-) But I think that most of people who answered to my first thread did not read well that release note I’ve linked to it or, almost, they did not read it at all!
The doc very clearly says: with DOT 7.3.1, on a 2020 (and only here), if you want bypass the 8 TB aggregate limit and have a 16TB one (as in other FAS!) you MUST configure a separate aggregate to host ONLY the root volume! Where the unpractical nonsense is? Is that if you have a base 2020 with 12 1 TB SATA and another shelf with 1 TB SATA disks, let say with only other 8 disk (total 20) we will have:
aggr0 - 3 disks RAID-DP with ONLY the root flexvol aggr1 – 9 disks from the base 2020 plus 7 from the shelf, raid-dp, 16 TB raw aggregate with all the other flexvols with user data one spare disk
First aggregate result: 3 TB raw just to host the default minimum size of root volume (roughly 12 GB on a 2020 if I well remember)! This is a nonsense waste! So we cannot, as possible in other systems, use only a couple of parity disk spanning this big aggregate, but we need to couples of that for the root need its own aggregate!
I suspect this is too do with the aggregate rebuild time on a 2020 as it will be pretty high with such a lowly processor. The risk of having a failed root aggregate for so long is probably too high for netapp to to accept it on your behalf.