So, I assign Filer-1 (Controller-1) 24 Disks from Shelf - 1 and 24 Disks from Shelf - 2 to Filer (Controller) 2
First we create a Raid Group,
A) Now we will create a RAIDGROUP (rg0) or (rg1) or can I name it (rg<something>), so I create a RAID-DP, 2 Disks for Parity, 1 HOT SPARE Per Controller, Per Disk type, in my case all SAS Disks, so 1 SAS Spare on each Controller (FILER).
B) On top of that RAID Group, I will create an "Aggregate", so now as I have 21 Usable Disks (Space), I will create TWO (2) Aggregates out of these 21 Disks or shall I create one large aggregate ?
C) On top of that Aggregate, we will create VOLUMES, which we will basically use for hosting File shares or Exchange Database or VMWare ?
D) On Top of Volumes, we create "QTrees", QTrees are like Logical partitions as in Windows we have C:, 😧 ?
So, if a RaidGroup (rg1) is created consuming all 23 Disks with 1 Hot Spare on First Shelf, All Aggregates created there after on top will be member of the same RAIDGroup as (rg1)?
Q-1 : What are the limitations, I mean how many Disks Does NETAPP recommends to allocate in a RAIDGROUP or the Maximum Number of Disks that can be part of a RaidGroup on NetAPP?
Q-2 : 64-BIT Aggregates can be more than 16TB in Size? Any Best practice or Recommendation for Aggregates ?
Q-3 : Maximum Number of Disks a Controller (Filer) can Handle? For e.g. if in future, we add another 2 Shelves, can we simply add those 2 more shelves to just one of the Controller (Filer) we have ?
Will be grateful if some netapp expert can have a look at my notes above, this will help me clarify my concepts prior getting trained on the product itself, this is the out-come of self learning, foreseeing our executives investing in NetAPP.
Thank you so much for your replies - it has been helpful, I got busy and hence the delay in responding back here, I will post some more questions either on this thread or will create a new thread - Once again thanks so much for your replies.
Both Controllers (FILERS) Keep a SYNC of each others configuration to take over in-case of a Filer (Controller) Failure ?
Well, not exactly - if one controller fails, its virtual identity is run inside of the surviving partner. You can actually misconfigure a number of things (apply different settings to two controllers, e.g. different VLANs, etc.), so your cluster fail-over ends up not very nicely. There is a tool for checking the cluster setup though: http://now.netapp.com/NOW/download/tools/cf_config_check/
On Top of Volumes, we create "QTrees", QTrees are like Logical partitions as in Windows we have C:, D: ?
Nope - qtrees are subfolders of a NetApp volume, so you can have some files inside a qtree; or you can have a LUN inside a qtree (or inside the root of the volume for that matter)
Any Best practice or Recommendation for Aggregates ?
If you have access to Field Portal, Storage Subsystem FAQ your best doc to answer these types of questions:
Most of your understanding is correct. To clarify, an aggregate is made up of one or more raid groups. You do not need to worry about creating raid groups. You create the aggregate and set the raid group size for that aggregate and as you add disks, Ontap will manage the underlying raid groups for you.