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

by Dianne McAdam
October 1, 2004

Network Appliance implemented a similar technology in its filers. Write Anywhere File Layout (WAFL) never overwrites old data with new data; it doesn’t need to copy the old data to preserve the snapshot of the volume. Similar to the SVA, NetApp filers create logical copies (also called snapshots) by making a duplicate set of pointers that point to the same physical location as the original volume. Since the filers don’t update in place, the snapshot continues to point to the old volume while the updated data is written to a new location. Neither NetApp’s nor STK’s logical copy implementations requires additional capacity initially to make volume copies.

Other vendors offer logical copy functions that require users to create separate areas within the array to save copies of older data. This type of process (often referred to as “copy-on-firstwrites,” “copy-on-write,” or “copy-out”) requires additional physical disk capacity as changes are made to the original volume. However, data is copied only from the original volume when a block of data is updated the first time.

Copy-on-first-write requires two writes for every update. As changes are made, the data is copied to a “save area” (the first write). Next, the original volume is updated (the second write). This compares to the snapshot functions supported by NetApp and STK in which a second write isn’t required since data isn’t moved to a save area.

How Copy-On-First-Write Works

Consider a production data volume (Volume ABC), which consists of only 10 data blocks. An array storage controller maintains an index for ABC, which points to each of ABC’s 10 data blocks. When a logical volume copy is created, no additional capacity is required. In fact, no data is copied because only a copy of the index of ABC is made, not a copy of ABC’s data.

The index copy (Volume abc) behaves just like the original index of Volume ABC and contains exactly the same point- ers to each of the original 10 data blocks for Volume ABC (see Figure 3). Both indexes point to the same 10 data blocks. As a result, the logical copying process requires no additional capacity, and a volume copy is made readily available (within milliseconds) to other applications.





The logical copy (Volume abc) never needs additional capacity as long as original Volume ABC remains static. However, in transaction-oriented production environments, data volumes are constantly updated. If primary Volume ABC is updated, for example, the view data represented by the index Volume abc must be preserved as of the point in time that Volume abc was created (i.e., prior to the update of Volume ABC).