What kind of space management to use

The type of space management you should use depends on many factors, including your tolerance for out-of-space errors, whether you plan to overcommit your aggregates, and your rate of data overwrite.

The following table can help you determine which space management capabilities best suit your requirements.

Note: LUNs in this context refer to the LUNs that Data ONTAP serves to clients, not to the array LUNs used for storage on a storage array.

If...	Then use...	Typical usage	Notes
You want management simplicity	FlexVol volumes with a guarantee of volume	NAS file systems	This is the easiest option to administer. As long as you have sufficient free space in the volume, writes to any file in this volume will always succeed.
You need even more effective storage usage than file space reservation provides You use automatic space preservation and actively monitor available space on your aggregate and can take corrective action when needed You have space savings from deduplication and compression and want to use the free space made available Snapshot copies are short-lived Your rate of data overwrite is relatively predictable and low	FlexVol volumes with a guarantee of volume AND Reservations on for LUNs and files that require writes to succeed AND Fractional reserve < 100%	LUNs (with active space monitoring) Databases (with active space monitoring)	With fractional reserve <100%, it is possible to use up all available space, even with reservations on. Before enabling this option, be sure either that you can accept failed writes or that you have correctly calculated and anticipated storage and Snapshot copy usage.
You want to use thin provisioning You actively monitor available space on your aggregate and can take corrective action when needed You want to share free space at the aggregate level to increase overall storage utilization	FlexVol volumes with a guarantee of none	Storage providers who need to provide storage that they know will not immediately be used Storage providers who need to allow available space to be dynamically shared between volumes	With an overcommitted aggregate, writes can fail due to insufficient space.