We are in the process of scoping out how to setup our 2240... this will be our first NetApp appliance so I am unsure how to yield the useable storage needed for the project in question. It will be an active/passive setup, no shelf, just the 2240 with 24x3tb SATA drives. Ideally, we'd like the highest amount of storage possible from this setup. When using the NetApp Calculator, if I use a RG of 24, that yields just over 40TB which is great, but upon research here and Google, it seems that having a RG around 14/16 seems to be the best practice for performance.
My question is, since we are doing active/passive, is an RG of 24 ok? There is a possibility of adding shelves to this in around 12+ months so I'm unsure if that effects this initial setup or not.
By default with SATA the max number of disks is 20 but setting options raid.raid_dp.raidsize.override on allows up to 22. So if the other controller isn't being used you could have a single raid group of 18D+2P with 4 spares or 20D+2P with 2 spares on that controller... that is for 7-Mode. If Clustered Data ONTAP, a separate mroot aggregate is required so you would have 3 disks for root, then 19 drives available for the data aggregate if you want 2 spares (or 20 and 1 spare but you lose maintenance garage with 1 spare).
I'm not familiar with the calculator you're talking about - can you provide a link?
A RG size of 24, with 3TB drives, should give you just under 60TB usable (22 data drives*3TB*.9 WAFL overhead) - but you'd want a spare, so say RG size=23, 21+2, ~56TB.
You say active/passive - but then you say "just the 2240 with 24x3tb SATA drives". Do you have one controller or two? If you have two, you'll have to allocate 3 drives (1D+2P) to the inactive controller for vol0, which leaves you a possible RG size of 21 (19+2), which gives you just over 50TB on your active system, assuming you don't split out your vol0 there. But that doesn't leave you any spares. If you leave one spare per controller, you're left with about 46TB on your active system, with a RG size of 19 (17+2). You could manage with a single spare on you active system, and assign it over to your inactive system if you need to, whice gives you back on data drive.
One thing to consider is future expansion. Once you decide on a RG size, you want to grow the aggregate by multiples of that many drives, to avoid performance hits. For example, if you go with a RG size of 23, each time you add a shelf you add another RG plus a spare, which is good. But if you go with a RG size of 19, then you add another 24 drives, you've got 5 drives either to use as spares (too many and wasted space) or you end up with an incomplete RG, which leads to poor performance.
I've done RG sizes of 22 drives, but not on SATA. It also depends on your project requirements - IOPS vs. space vs. expansion vs. redundancy...
There are 2 controllers in our 2240, and yes, I did forget to mention the 1D+2P for vol0 on the passive controller, my fault. The main purpose of this setup is space. Performance isn't an issue since its mostly people doing runs locally on their workstation and offloading to the FAS, about 15 or so people maximum.
Okay, so assuming you don't split out vol0 on the active controller, and two spares, you have 19 drives on the active controller, which will give you ~45TB. I think your screenshot is assuming two seperate vol0s, which isn't necessary.
Your main focus is space, so you want a RG size of at least 19 drives - you can set it higher, but at this point it won't make any difference since you will only have 19 drives to work with. If your focus is still space when it comes time to buy more drives, you may decide to increase the RG size - though I'd have to check what NetApp recommends as the max. If you're not worried about performance due to partial raid groups, consider setting a RG size of 23. That way, you get a spare for each new full shelf.