A Consistent Point in Time

Over the last few years continuous data protection (CDP) has gone through the typical life cycle of a new technology. When it was first introduced, CDP was  heralded as the killer-app for backup software. Since it could provide real-time recoveries at any past point-in-time with minimal or no application down time, why use backup software?

Then questions and issues around CDP began to surface. How long could CDP products retain data? How do the products differ? Do they differ? How do CDP vendors avoid creating too much overhead on the host server. And what is CDP anyway? Frequent snapshots? Journaling? A combination of both? These unanswered questions led to CDP falling below the radar screen while the products matured and analysts and vendors alike sorted out the issues and products.

The good news is that this quiet period is drawing to a close and CDP is starting to make some noise again. CDP has arrived at a point in time where the best use scenarios for CDP products are largely defined and companies can break CDP products out into different categories. The bad news is that few people outside of the storage industry know what CDP means, much less how CDP products differ.

A definition for CDP can now be found on the SNIA web site. In short, it defines CDP as continuous data capture, storing data in a location separate from primary storage and providing the means to recover data to any past point-in-time. This definition should help to minimize the number of vendors that inappropriately apply the term CDP to their other versions of data protection software.

Yet the real story is that the story for CDP has changed. Providing simplified, near-real-time recoveries for servers across the enterprise with minimal administrative effort is a growing concern for organizations. Applications on one server often have hidden dependencies with applications on other servers that only a few people in companies really understand. Making the situation worse, no one can really document or explain what the impact is if one application fails. Further if one application generates inaccurate data, what is the damage should that bad data spread to other applications. How do companies recover back to a consistent point-in-time when they know the data is reliable?

Companies have come to a point that providing real-time recovery for just a few select servers or applications labeled as "mission-critical" is no longer good enough. Enterprises now have some many applications that are so dynamic that a specific application server can rise from the status of second-tier to mission-critical at any point in time.

This is why DCIG Inc sees the new generation of CDP products so valuable. Companies need to stop thinking parochially about recovering a single server to a specific point in time but should begin to think globally about recovering any of their enterprise application servers to a specific point in time. Once CDP takes on that perspective, corporate management should start to view CDP as good business practice rather than as just a point solution.