- All FAS systems
- All AFF systems
As ONTAP continues to push the limits of storage efficiency in every release, there are customers who stand to benefit from these innovations. For example, some systems might have their root aggregate on whole drives and customers may want to move to root-data (ADPv1) or root-data-data (ADPv2) partitioning to gain additional storage efficiencies, such as reduced parity and spare disk overhead.
Also, some customers may want to move from ADPv1 on HDDs to ADPv2 on SSDs to reap the benefits of Flash and further improvements in parity and spare disk overhead. Keeping that in mind, this article is written to cover the steps required to migrate the root aggregate from whole drives to ADPv1 (or ADPv2), as well as the steps to migrate the root aggregate from ADPv1 to ADPv2. Note that ADPv2 is supported only on SSDs.
This procedure is nondisruptive – customers can continue to run their applications while the root aggregate is migrated.
PrerequisiteTo move to ADPv1/ADPv2, the following are prerequisites for a successful migration:
- The nodes must be part of an HA-configured system (HA pair).
- You must be running an ONTAP version with the fix for BUG 935935 – Feature Request: The "system node migrate-root" command needs to support partitioned disks
- There must be at least eight drives of the drive type that one needs to move the root aggregate to. The recommended number of drives per node are 12 or more.
- There must be an even number of destination drives, so that those can be distributed equally between the two nodes of the HA pair.
- The destination drives must be spare disks and not already partitioned.
- The destination drives must be large enough to accommodate root aggregate slices on its thin partitions and still leave significant space for one or two data partitions
- The FAS or AFF system currently hosting the root aggregate must support ADP
- Entry-level platforms support ADPv1 on HDDs
- Mid-range platforms support ADPv1 on HDDs only via bootarg
- All platforms support ADPv2 on SSDs
- Both the nodes in the HA-pair must be in a healthy state with no RAID groups performing reconstruction, sick disk copy, or re-parity operations
- The root aggregate must not be mirrored on either node