This is part 2 in the Lost in Space series on how space allocation works in Data ONTAP. This post covers the physical space aspect.
Raw vs Usable – Physical
In this post, we’re deciphering how much space you bought versus how much space you actually have available to use. The focus here is on physical space, before we even get to the software portion of ONTAP.
When you buy a disk, you pay a dollar amount per space unit. These days, it’s $ per GB. In the future, we’ll probably pay $ per PB. Slower drives with higher capacity (such as SATA) will cost less per GB. Faster drives with lower capacity (such as SSD) will cost more per GB. The good thing about technology, however, is that prices normalize over time as production of new stuff gets cheaper.
The file system
What most people don’t think about is that when you buy a hard drive, you are paying for what is considered raw space. The hard drive is useless until it is formatted with a file system. This is true of *any* storage device, whether it’s a Windows desktop using NTFS or an All-Flash FAS.
NetApp’s WAFL file system takes up ~10% of the raw drive space for its file system. NTFS takes up a minimum of 12.5%. Other storage vendors use varying amount of raw space for the file system, features, etc. The 10% is not modifiable.
Hard drives in enterprise storage (or even servers) will also often participate in some sort of virtualized aggregation of disks to allow for better performance, redundancy, etc, generally known as RAID. In Data ONTAP, a default of RAID-DP (a double-parity variation of RAID6) is used, with the option of using RAID4.
What that means in terms of disk is this – Every time you create a group of disks for use with your storage system in ONTAP (known as a RAID group), you will lose at most, 2 disks for parity. That eats into your “usable” calculations.
RAID groups are contained within aggregates. An aggregate can have multiple RAID groups. These RAID groups can vary in size. The size of the RAID group will depend on what types of disks you have, what the intended workloads are, etc.
How much usable physical space do I really have?
Your usable disk space is considered to be:
(Number of disks – number of parity disks) x HDD raw space = Raw space adjusted for RAID
Raw space adjusted for RAID – (Raw space adjusted for RAID * 10%) = Total usable
If you have 2 RAID groups with 10 disks each and use RAID-DP, you essentially have 16 total disks for actual data use. 4 disks (2 for each RAID group) are dedicated to parity disks.
If those disks are 1TB each, that means you have 16TB of raw physical space.
Once you format the disks for use with WAFL, you then have 16TB – (16TB * 10%) = 14.4TB of usable physical space.
What’s important about physical space?
The key thing to remember about physical space is that no matter how much space your OS tells you you have available in a volume, share, LUN, etc. you are always going to be limited to the physical amount of space available in your system.
Future posts will cover the virtual aspect of storage space allocation.
That about covers the physical space aspect of space allocation in Data ONTAP. If you have questions or want me to add something to this post, feel free to comment!
Stay tuned for more about space allocation in this series!