Deployment

The Cloud Object Storage manager provides Cloud Object Storage operators the ability to configure and monitor Replication operations within the Cloud Object Storage.

This configuration can be found on the page within the manager. This configuration spans storage-pools and container vaults. To use Replication, the feature must first be enabled at the storage-pool level and then at the container vault level.

When configuring Replication onto the storage-pool, operators must configure a rate-limit for the synchronization operations occurring within the storage pool. This rate-limit controls the amount of load that is generated by the background process that executes Replication work items that are found within the Replication work item queue. If this rate is set too low relative to the amount of Replication work generated within the storage-pool, then this introduces latency between work item creation and execution. If set too high, then work item execution may degrade performance of user-initiated S3 operations within the storage-pool. Replication sync latency observability is provided to help making rate-limit tuning decisions.

Operators have the option of disabling replication on vaults. This prevents users from enabling new replication policies on their buckets and pause background replications for any existing policies inside the vault. When background sync is paused, users’ replication work will continue to accumulate in their respective queues, and their processing resumes once background sync is re-enabled.

Once Replication is enabled at the storage-pool level, operators can then enable Replication on container vaults that are deployed to the storage-pool. This configuration can be found on the page within the manager. Once enabled, Cloud Object Storage users with S3 access to buckets that are located within the container vault has the ability to set Replication policies on their buckets. These policies are allowed to target any bucket that is housed in a container vault that also has replication enabled.

Operators have the option to specify the S3 endpoint that is utilized for servicing object synchronization requests targeting buckets that are located within the container vault, including those originating from within the container vault. By default, Accesser® devices executing synchronization work items address their synchronization requests to the IP addresses of Accesser devices that are deployed to the container vault that contains the target bucket. The ability to override this behavior is intended for complex Cloud Object Storage deployments wherein direct network connectivity does not exist between Accesser devices that are deployed to replication-enabled container vaults. The endpoint that is provided can be either a hostname or an IP address that provides access to the S3 API for containers located within the container vault. A port configuration is not allowed as replication synchronization requests will always be HTTPS over port 443. When setting this override the operator should help ensure that all Accesser devices that are deployed to all replication-enabled container vaults have network connectivity to the provided endpoint.