HPE Storage Users Group

I'm trying to work out what and how space is used on my 3PAR.
eg
I have 8936GB of Total Raw FC capacity.
I've created a 1024GB TPVV for some test VMs, this VV shows a Virtual Size of 1024GB (so far so good), a Used User Size of 551GB (TP at work), but if I look at the Physical Disk Allocated Capacity, it says 3362GB, with Free Capacity of 5574GB.
I expect a bit of overhead, but 3362GB of raw disk being consumed for a 1024GB VV is a bit rough?

Is there a way to approximate how much raw disk should be used? I realise there are a lot of variables, but even a ballpark formula would be good. I'm using mostly default options (R5 3+1), copy space is selected, does this effectively double the allocated raw space?


Edit:
I think I've made sense of it. I believe the "spares" are allocated at the beginning. So because I have 8936GB Raw, the CPG (?) has reserved 2234GB for spares, and 6702GB for usable (ie 3+1).

Got to get out of the EVA mindset and more into 3PAR

Ok more questions:

1. On my FC CPG (R5 3+1) I have 8936GB Raw (total capacity) and the spares add up to 2234GB which perfectly works out to 8936/4.
But on my NL CPG (also R5 3+1) I have 29800GB Raw, but my spares only add up to 5216GB (This figure doesn't work out to any integer ratio of 29800). But if I add 5216 to the 2234GB above, it equals 7450GB which does work out perfectly to 29800/4.
I hope that makes sense? Does this mean the NL is sharing Spares with the FC? Shouldn't I have 7450GB of dedicated spares for my NL?

2. Next up. On my FC, I've provisioned 2708GB of VV's which shows up as a Virtual Size of 2708GB, but under the Physical Disk disk it says the "Volumes" total 2025GB. Compare this with my NL , I've provisioned 18432GB of VV's, but "Volumes" says a total of 23163GB. How does "Virtual Size" and "Volumes" relate? I thought they should be the same number?
In my case one is higher and one is lower which isn't helping to figure it out. All Volumes are TPVV. The main concern is the NL which is using 5TB more than I've allocated, and now I'm running out of space.

Ok I've figured a bit more out...

The "spares" are separate from the RAID overhead (why?), so you get stung twice. On my FC, I have 4 disks worth of spares (out of 16 total), and in my NL have just under 3 disks worth (also 16 disks total - how are these spares this calculated?). Then when I provision a VV, it takes another 33% (R5 3+1) for RAID overhead. Then there's some admin space too etc.
So overall, for my 9600GB (16x600GB) of FC that I bought, I get 5026GB of user usable space (52% of total). 3PAR is good, but at 52% usable it has just doubled in price!

This is something more noticeable on low spindle counts, from what I've seen it automatically reserves around two physical disks worth from each disk type (NL/FC/SD) in case of disk failure. This is before you create any CPGs and pick raid type/size. Then you get some admin space and beyond that what ever parity etc from CPG settings.
With 100+ spindles it's less noticeable but as I discovered recently with only 16 * NL2000 I was left wondering where my 4TB went?

My understanding of the reasoning behind "spare" + "raid overhead" is that you then have space to perform a raid rebuild without waiting for somebody to physically replace the disk.

If you just have the "raid overhead" then in the case of a disk failure you end up running degraded until such time as the disk is replaced.

Re: Spares

My system has 384 FC spindles of 300g drives, 107 TB total, with 2,787 gb of spare. Thats about 10 drives worth of spare space. Each magazine holds 4 drives.

For NL, I have 96 spindles of 1 TB drives, 89 TB total, and only 936 gb of spare. Thats only 1 drive's worth of NL spare. Each magazine still holds 4 drives.

Based on that, I would have to concur that FC space can be used to hot spare NL drives.

Regarding getting stung twice, that has always been the price for hot spares with any raid implementation. Traditional raid arrays (like Netapp, Clariion, etc) dedicate spares on a per drive basis... so with 4 hot spares, you have 4 spindles idle doing nothing for the performance of your system. With 3PAR (XIV, Compellent), spare space is distributed through out all the drives in the system, so that every spindle installed can be used for maximum performance. When a 3PAR drive fails, the rebuild is distributed accross all the spindles/shelves for faster reovery times. Traditional arrays rebuild times could be several hours depending on the size of the LUN being re-written.

Thanks all