Volume pinning is a storage optimization strategy available in clusters with tiered storage (including both hybrid and all-flash configurations).
By default, data with a higher access frequency in the volume will remain in the cache layer, which has better processing performance, while data with a lower access frequency will move down to the data layer. In practical scenarios, some applications might require all data to remain in the cache layer at all times to provide high-performance services. In this case, you can enable volume pinning for the volumes of these applications, ensuring that the data is always retained in the cache layer.
Prerequisite
The cluster must have tiered storage enabled.
Information
The cache reserved for pinning refers to the total cache partition capacity reserved for the pinned volumes. The calculation formula is as follows: Cache reserved for pinning = (Total write cache capacity - Invalid write cache capacity) × Reserved cache ratio. Up to 99% of the reserved cache capacity on each node is available for storing volume pinning data. If the data size for volume pinning exceeds 99% of the reserved capacity, the placement requests will be rejected. It is recommended that the cache reserved for the volumes to be pinned to be 10% to 20% larger than the required capacity.
The maximum capacity that can be allocated to a single pinned volume depends on the total cache capacity reserved for pinning and the redundancy policy settings of the cluster.
Maximum capacity allocated per pinned volume = Total cache capacity reserved for pinning in the cluster / Replication factor = Total cache capacity in the cluster * 0.8 * X% / Replication factor
The replication factor is determined by the following rules:
If the volume uses replication as the redundancy policy, the replication factor is as specified in the redundancy policy.
If the volume uses erasure coding as the redundancy policy, the replication factor depends on the erasure coding configuration:
Precaution
When no file storage cluster is deployed in the cluster, and there are no virtual volumes, iSCSI LUNs, or NFS files with volume pinning enabled, you can disable Allow volume pinning. Once this option is disabled, you can no longer enable volume pinning for virtual volumes, iSCSI LUNs, or NFS files in the cluster.
Procedure
On the AOC cluster page, click a cluster, then select Settings from the tab bar, and select Volume pinning.
Enable Allow volume pinning.
Specify the percentage of the physical disk pools' write cache capacity that can be reserved for pinning. You can set a default reserved cache ratio for all physical disk pools, or set it individually for each physical disk pool.
Set a default reserved cache ratio
Set a default reserved cache ratio for all current hosts in the cluster and any new physical disk pools added later. The ratio can be set from 1% to 75%. If you tick Apply to the physical pools of all hosts, all current hosts in the cluster will reserve cache for pinning based on this ratio.
Set reserved cache ratio for physical disk pools on specific hosts
If you want to place the pinned volumes only on physical disk pools on specific hosts, or set different reserved cache ratios for the physical disk pool on each host, you can untick the Apply to the physical pools of all hosts checkbox, and then set the reserved cache ratio for the physical disk pool on each host individually.
Note:
- The reserved cache ratio for a single physical disk pool can be 0%, which means that this pool will not reserve any cache for volume pinning. If the physical disk pool contains data partitions, the maximum reserved cache ratio can be set to 75%. If the physical disk pool does not contain data partitions, the maximum reserved cache ratio can be set to 100%.
- If reducing a physical disk pool's reserved cache capacity would cause the cluster's total reserved cache to fall below the total cache currently used by current pinned volumes, lowering the physical disk pool's reserved cache ratio is not allowed.
- For clusters without the active-active clustering feature enabled, you need to set a reserved cache ratio for the physical disk pools on at least three hosts in the cluster. For active-active clusters, you need to set the reserved cache ratio for the physical disk pools on at least two hosts in each availability zone.
Click Save to complete the configuration.