Other best practices for StorageGRID and FabricPool
When configuring a StorageGRID system for use with FabricPool, you might need to change other StorageGRID options. Before changing a global setting, consider how the change will affect other S3 applications.
Audit message and log destinations
FabricPool workloads often have a high rate of read operations, which can generate a high volume of audit messages.
-
If you don't require a record of client read operations for FabricPool or any other S3 application, optionally go to CONFIGURATION > Monitoring > Audit and syslog server. Change the Client Reads setting to Error to decrease the number of audit messages recorded in the audit log. See Configure audit messages and log destinations for details.
-
If you have a large grid, use multiple types of S3 applications, or want to retain all audit data, configure an external syslog server and save audit information remotely. Using an external server minimizes the performance impact of audit message logging without reducing the completeness of of audit data. See Considerations for external syslog server for details.
Object encryption
When configuring StorageGRID, you can optionally enable the global option for stored object encryption if data encryption is required for other StorageGRID clients. The data that is tiered from FabricPool to StorageGRID is already encrypted, so enabling the StorageGRID setting is not required. Client-side encryption keys are owned by ONTAP.
Object compression
When configuring StorageGRID, don't enable the global option to compress stored objects. The data that is tiered from FabricPool to StorageGRID is already compressed. Using the StorageGRID option will not further reduce an object's size.
Bucket consistency level
For FabricPool buckets, the recommended bucket consistency level is Read-after-new-write, which is the default setting for a new bucket. Don't edit FabricPool buckets to use Available or any other consistency level.
FabricPool tiering
If a StorageGRID node uses storage assigned from a NetApp ONTAP system, confirm that the volume does not have a FabricPool tiering policy enabled. For example, if a StorageGRID node is running on a VMware host, ensure the volume backing the datastore for the StorageGRID node does not have a FabricPool tiering policy enabled. Disabling FabricPool tiering for volumes used with StorageGRID nodes simplifies troubleshooting and storage operations.
Never use FabricPool to tier any data related to StorageGRID back to StorageGRID itself. Tiering StorageGRID data back to StorageGRID increases troubleshooting and operational complexity. |