Snapshots use significantly less space and are more efficient than data mirrors. A mirror is an up-to-date copy of data for a logical volume. Two or more complete copies can exist at the same time, although only one copy is seen or used by an application, so mirrors require double or more the amount of disk space than the original data.
A snapshot is a view of data at a specific point in time, much like a photographic image is a snapshot of physical images at a particular point in time. You can use snapshots to validate data before you save it to permanent storage, data mine and generate reports, and retrieve specific data items.
Snapshots are stored in a different location than the replica so that the replica can continue to march along in time. The snapshot, however, is frozen with respect to the replica. Again, using the analogy of a photograph, you can now draw on the photograph and it does not effect the original subject of the photograph. The ability to modify the snapshot is accomplished by using a copy-on-write log file.
Notice from the above figure that data is passing through the datatap on the recovery server in the case of reads and writes to snapshot data. Assure MIMIX DR for AIX uses a different set of device minor numbers when dealing with snapshots, so that the datatap knows which log files to access in a specific order. For example, when a write operation is directed at the snapshot it is actually written to the copy-on-write (COW) log instead. If the data has not been modified, then a read operation would come from the snapshot. If the data has been modified, then the read would come from the copy-on-write log. Keep in mind that the snapshot is the representation of the application data at a specific point in time.
Related Topics: