Which Configuration is better for performance ?

[AllenChang]

If storage have 24 NVMe SSD,

Customer want' using CIFS Prototol, and Every controller have two 10Gb SFP+ be their interface,

I am consider two solution

1. Use all the 24 disk partition to crate one Aggregate, and one flexVolume, and create a SVM  that have four lif,every controller have two lif,and use DNS Run roubin to load sharing

2.create two aggregate,and one FlexGroup accross the two aggregate, and create a SVM  that have four lif,every controller have two lif,and use DNS Run roubin to load sharing

Which will better in performance?

[SpindleNinja · Accepted solution]

The second cause the first doesn't really make sense on an AFF system, IMHO.  although people do it.    Only real reason to do it is if you need a single giant lun that exceeds the capacity of a single aggr. 

And by using Flexgroups you'd be able to use both controllers, as well as multiple processes across each controller (# depends on how many member volumes the FG has.