Exploring Storage Snapshot, Cloning & Copy Techniques: Is It a Logical or Physical Copy?

by Dianne McAdam
October 1, 2004

While all the leading disk-array families support volume cloning, not all deliver the same level of functionality. For example, some products let IT administrators create more than one clone at a time or access a clone before the copy operation finishes; others let users create a clone of a clone. Some arrays support cloning for both z/OS and open systems volumes, while others support only open systems volumes.

When evaluating cloning software, IT administrators should ask the following questions to determine the suitability of the product to the given application:

• How long does the cloning process take? For example, assume that 50GB of data has to be copied from one volume to another. The time it takes to complete the cloning process differs from vendor to vendor and array model to array model. Completion time can also vary widely when subsystems are heavily loaded and when multiple concurrent copy operations run. Cloning large volumes can take hours, even days, to complete.
• Can a clone be used before the copy operation is complete? Some implementations allow clones to be mounted by another application before the copy operation is completed. In these situations, the copy operation continues to run in the background. This feature can be a big time-saver when cloning large volumes.
• How many clones can be created in a single array? Some arrays restrict the total number of clone copies; others place restrictions only on the physical capacity and number of volumes the array supports.
• How many clones can be created of a specific volume? Some implementations restrict administrators to one clone per volume, while others allow up to 12 or more clones per initial volume. Creating multiple clones of the same data allows multiple processes to occur concurrently. For example, one clone can be used to run a backup, a second can be used for testing purposes, and a third can be used for data warehousing.
• Can clones be created from clones? Though this may sound like a redundant feature with no benefit, it can be useful for developers looking to test new versions of applications. It ensures that the test volume remains intact in the event of a “bug” or any unwanted changes. If the test volume becomes corrupt or is altered in any unwanted fashion, the second clone can be used for continued testing.
• After the one-to-one relationship between the original volume and clone is broken, how long will it take to reestablish the relationship so the clone is a mirror image of the original? To reestablish this relationship, will the controller copy make a full, new copy of the original data (in effect, creating a whole new clone) or will it copy over only the data that has changed since the relationship was severed? Copying only the changed data saves time and processing cycles.
• Can the controller restore a damaged or corrupted production volume using the clone as a source? If so, does the controller copy all the data or only the changes since the relationship was severed? One of the best uses for clones is making point-in-time copies of critical data. If you must restore from a clone, you want the process to occur as quickly as possible.
• How easy is it to configure a clone? Can clone volumes be created at will or does a pre-configured pool of empty volumes have to be created first? Can IT administrators reconfigure the pool or is this task left to the vendor? The easier it is to create clone volumes, the less time IT administrators must spend on the task. However, care must be given to tracking and administering the clone volumes once they’re created.
• Are all clones RAID-protected? Can the Redundant Array of Inexpensive Disk (RAID) groups containing the clones be automatically repaired through dynamic sparing? If the clones aren’t RAID-protected, a hardware disk failure can result in losing the cloned data. While this level of protection may seem redundant, it can be critical. An unprotected clone can disrupt applications if there’s a disk failure (it can disrupt an application that happens to be using a clone on the failed disk). For example, it can cause IT administrators to have to create new clones and restart entire backup jobs.
• What operating systems are supported? If your environment includes both mainframe and Unix servers, can you create copies of z/OS volumes and open-systems Logical Unit Numbers (LUNs)?
• How are the clones managed? Does the vendor provide management packages that help IT administrators determine which volumes are clones and their status? Are the clones still established with the original LUN or volume, or have they been split? It’s important to constantly manage all the clones within an array to ensure that extraneous clones are deleted and capacity is freed for other purposes. Getting the right answers to these questions can help you understand the capabilities and limitations of the cloning function.

Vendors That Create Physical Clones