I need to add 2x24 SAS shelve into an existing aggregate, and it's current layout as shown in the attachment.
I am going to create two more raid groups with the same size of 22 disks as rg4 and rg5 in the same aggr. However, my questions is, should I also need to use extra 2 disks into rg3 to make it up as full raid group? would that cause hot spot on these two disks? ontap 8.3.2 cdot.
Yes , I would certainly worry about the hot spots after adding these two new disks to rg3 .
I would "reallocate" volumes after adding disks.
Reallocate has to be performed whenever you add new disks to aggregate or if your aggregate is 80% full and if you have "hot spots"
"reallocate" volumes would take a lot of time and efforts on all volumes, so, if it could really happen, I probably should sacrifice the opportunity to add 2 disks (~2TB). Agree?
Also, can somebody please tell me will these two new disks equally take incoming data along with 2 new raid group, if yes, then there should not be any hotspot, because there will be 2x22 +2 = 46 disks, enough spindles to stripe out data. Am I right?
Two New disks to rg2 will not equally take data as other two raid groups. Hot spots will occur and the only solution is to "rebalance".
You have to sacrifice adding disks to rg2 if you consider "rebalancing" is going to casue performance impact.
This link below indicated the data should be stripped out across all new added disks, independet of raid group, if true, then in my case, these two new disks will be taking the data equally along with two same time new added raid groups, which should have enough spindle to balance out data, therefore, there should be no hotspot.
This link does not explicilty talk about the data written to raid-groups, I am sure two disks that you add to rg2 will be utilized, but as the best practice particularly for rg2, reallocating will help you avoid "hotspots". I am not saying that they will not take data equally. They will, but balancing the blocks on the drives for a raid-group is important to avoid hot spots.
Flexible volume will stripe its data across all disks in the raidgroup. Manually reallocating will help you to make use of all the new drives in the raid-group are equally used.