We are coming up with a Getting started course on how this selection is done when more than one aggr is eligible.
Least-Utilized (only option prior to 5.1)
–Selects aggregates with most available space
–Margin can be tuned to select acceptable range
Best-Fit (new with 5.1)
–Selects aggregates with the least available free space that would fulfill the need
–Margin can be tuned to select acceptable range
¡Global setting to select preferred algorithm
Manually resource selecting ignores both algorithms
There are two phases namely space balancing and load balancing.
In space balancing phase we pick the aggr with largest free space( if least used criterion which is default) and all those within 75% of its capacity.
By this we come up with a set of aggr, and on this set we apply the space balancing like number of qtree or relationships.
In case of Best fit criterion we take the aggr that just fit and all those with in 1.75x its capacity and apply the load balancing.
In case of Primary provisioning the number of qtree is used and in case of secondary provisioning we use the number of relationships.
In the space balancing phase we select the aggr with largest free space and apply
Regards
adai