Posey on CDP: Continuous data protection catches on as backup strategyDate: Jan 24, 2013
A few years ago, organizations relied on tape backup to store their data, and continuous data protection (CDP) was a potentially pricey, complex and buggy alternative. But times have changed for CDP; continuous data protection is coming into its own. In this Storage Decisions video, backup expert and seven-time Microsoft MVP Brien Posey digs deep into the pros and cons of continuous data protection.
"The basic philosophy behind it is that it's better to back up data on an ongoing basis, usually at the storage block level, rather than to make one big monolithic backup each night. Essentially, if a storage block is modified, it gets backed up. Simple as that," Posey said.
The technology backs up changes to data as they occur and "totally eliminates the problem of an ever-shrinking backup window," Posey said. "With continuous data protection, there is no backup window. You're backing up data continuously throughout the day rather than trying to cram it into a backup window late at night. Also, open files are no longer a concern, because we're not even dealing with files anymore; we're dealing with storage blocks."
Posey said that when CDP first became available around 2006 or 2007, organizations were typically performing tape-based backups. He said that CDP -- which is disk-based -- required a "fairly substantial" investment in storage equipment.
"A lot of organizations didn't want to make that investment," said Posey, who later noted that there was a learning curve for implementing CDP and that there were other issues with CDP early on.
"Operating system compatibility and application awareness were huge issues, and to be perfectly frank, some of the early continuous data protection solutions tended to be pretty buggy," Posey said.
"In terms of asynchronous, the term 'continuous data protection' is a little bit of a misnomer. When you back up your data, it is not truly continuous. Instead, asynchronous continuous data protection products use a synchronization schedule. Generally, this schedule is every 15 minutes, but it can be custom-tailored," Posey said. "So rather than constantly springing data to the server in real time, it does get staggered out according to the replication schedule."
Synchronous CDP can mean data is first written to the backup, and then to the file or database server to ensure that the data is protected, he said. "Synchronous tends to be a lot more expensive and is intended for use in organizations where absolutely no data loss is tolerable -- things [such as] financial organizations, where they can't stand to lose even one byte of data," Posey said.