(866) 683-8844 Phone
(609) 683-9664 Fax
338 Wall Street
Princeton, NJ 08540
(Directions)
This
page is
printer friendly.
TechNote:
Continuous Data Protection
Continuous Data Protection (CDP) Defined
Continuous Data Protection (CDP) is a process by which a data storage set is backed up as it is modified. The data is stored in a manner that allows the restoration of the original data source as it appeared at any previous point in time. The term "Continuous Data Protection" has been somewhat inconsistently adopted by a number of vendors to describe a wide range of capabilities and approaches to CDP.
CDP technologies are complex, and must be application and file-system aware in order for the recovery of the data to be possible at any point in time. Some CDP approaches compromise on the granularity of the recovery. For example, recovery can only be possible from periodic snapshots of the data. This may, in fact, be sufficient for many applications, but the choice of the approach to CDP must be made with these distinctions in mind.
To allow for a full recovery of a volume a time zero image of the volume must be completed before a CDP system can move to the "ready" state. CDP systems generally have a finite amount of storage, and at some point a CDP product must discard stored data. To maintain the backup integrity a new time zero image must be recreated.
There are three common types of CDP products: Block based, File based and Transaction based.
Block based systems back up each block on a storage volume as that block is changed. To recover data using a block based CDP an entire volume must be restored to a given point in time. Since a block based CDP application has no knowledge of the data it is handling, there is no way for the stored data to be stored for searching. However, since a block based CDP system does not interpret the data it is handling, it can backup any data volume or application.
Block based CDP storage volumes may not be in a consistent state, and there will be points in time when the stored data is unusable. The recovery process may require several recovery attempts from successively earlier points in time before a consistent data image can be obtained.
File based systems backup files as they are written. In this kind of system individual files may be recovered. Many CDP products (such as Storactive's LiveBackup for workstations) now allow individual users to recover their own files. Most file based CDP products allow the use of policies to control which files get backed up and for how long.
A file-based system can index for later search the files it can interpret. When this is done a file could be recovered by some part of its content as well as it's name, location or modification data. File based CDP systems have a tendency to not handle applications that continually hold files open. Mail systems and database packages are common examples of these.
Transaction based CDP products store changes to records stored in a database. The most common form of this is an Email system, and Storactive's LiveServ product ensures that mail messages and attachments are backed up as they are sent and received. Data recovery can be accomplished by the end-user from any point in time.
Each protected application requires a different CDP product that can handle its specific API and data properties. Since a transaction based CDP product is specific to a given application, indexing and other search and recovery features can be custom tailored to specific regulatory and customer requirements.
Key Point Summary
Continuous Data Protection:
- Block-Based: Volume level recoveries only. Data may not be always in a consistent state.
- File-Based: Allows recovery of individual files, and depending on policies, a number of prior versions of files.
- Transaction-Based: CDP utility must be application aware to capture discrete events, such as record changes in a database