A mirror is a read-only copy of an entire volume. Unlike snapshots, which are point-in-time images, mirrors are full read-only copies that can be kept synchronized with the source volume. Mirroring, which was previously impossible with Hadoop, lets you protect and manage data in several new ways:
- Isolation - keep production data intact while letting a research team work with a separate read-only copy
- Disaster recovery - mirror important data to a geographically separate cluster in another datacenter to provide continuity in the event of a disaster
- Load balancing - create multiple mirrors within a cluster and load balance among them to parallelize high read volumes
Using the MapR Control System, you can quickly create a mirror of any volume on the cluster:
- Click the New Volume button to display the New Volume dialog.
- For the Volume Type, select Mirror Volume.
- Type the name of the original volume you created in the Mirror Source Volume field, followed by the @ symbol and the name of the cluster ("default"). For example, if you have created a volume called
- In the Volume Name field, type a name for the mirror volume. Example:
- In the Mount Path field, type a place to mount the volume. Example:
- Click OK to create the mirror volume.