Storage Foundation Cluster File System High Availability 7.4 Administrator's Guide - Linux

How does InfoScale migrate data using cloud connectors

InfoScale supports block-level data migration to cloud. It uses the capabilities of VxFS multi-volume file system and VxVM volume sets to provide a stable infrastructure to connect on-premise and cloud environments. The cloud storage is represented as a volume in a disk group. A single file system is created over a set of volumes comprising a group of local volumes and cloud volumes. The multi-volume file system is mounted on the same mount point that was in use by the existing data volume. This allows applications to seamlessly access data even after data is moved to the cloud. A cloud volume is a regular VxVM volume with vxcloud attribute enabled. This volume must be manually tagged as a dataonly volume to avoid file metadata from being written onto it. Based on the attribute setting, VxVM decides whether storage must be provisioned locally or in the cloud. The cloud volume is not limited by locally available storage capacity. Each cloud volume is associated with a target storage unit called buckets or containers (the terminology varies with the vendor). The local and cloud volumes are assigned tiers across which files can be migrated. All I/O requests on cloud volumes are managed by the vxcloud daemon. The migration is based on the policy file, policy.xml, which defines the storage placement policies for your data. The policy file can be customized to suit your needs. The migration can include regular files; not empty directories or symbolic links.

In case of block-level migration, data is stored in the cloud volume in blocks of fixed sizes, each block representing an object. The object can hold data from different files. The object size corresponds to the block size of the file system on the local volume. For example, a file of 10 KB stored on the local volume with a file system block size of 2k will be written to the cloud volume in 2k block sizes as 5 distinct objects.

In both the migration types, each volume in the volume set is assigned one of the following user-defined placement classes:

File movement across the local and cloud tiers is defined by the policies assigned to and enforced on the file systems.

For more information on the policy file, placement policies, and rules:

See the chapter Administering SmartTier in this document.

Figure: How InfoScale migrates data using cloud connectors illustrates the process.

Figure: How InfoScale migrates data using cloud connectors