Continuous data protection (CDP) represents the future of data backup as companies look to provide near real-time recovery points to enterprise applications. Yet CDP software is still in various stages of ramping up, while near-CDP (frequent snapshots) software is already mature and available on many operating and storage systems. To help companies choose between CDP and near-CDP software, here's a checklist of questions to help you determine if you really need CDP.
What's the application's recovery point objective (RPO)?
CDP products can typically deliver recovery points to within seconds of an application failure or corruption. The best that most near-CDP products support is the creation of snapshots approximately once every 15 minutes. Scheduling snapshots to occur once every hour, once every eight hours, at the beginning, middle and end of the business day, or once a day are other common near-CDP settings.
What's the application's recovery time objective (RTO)?
The length of time to recover data is dependent on the scope of the outage, the amount of data to recover and whether it's recovered to a local or remote site. To locally recover small amounts of data (1 GB or less), recovery times for CDP and near-CDP are comparable -- 30 minutes or fewer. However, as the amount of data and scope of recovery grows, recovery times can also grow. As a rule of thumb, assume recovery times of an hour or longer using near-CDP, while CDP software can recover data in 30 minutes or fewer.
What are the write I/O characteristics of the application?
Copying write I/Os of high-performance applications may slow application performance while writes are copied, may require high performance Fibre Channel or SAS disk drives and storage systems to keep up with high application write I/O throughput. Using near-CDP, the application is only interrupted momentarily when a snapshot is taken and only requires as much disk space to track changed blocks.
How long do you need to retain the data for recovery?
CDP software can require a great deal of storage to keep all copies of the application write I/Os. Though there's no theoretical limit to how much data CDP software can retain, many companies have to delete data after it's 30 days old, and data from high-performance applications may have to be deleted as soon as three days after the data is created. Data can be retained for much longer periods of time with near-CDP software, but near-CDP software typically has an upper limit of how many snapshots it can create and manage. Though the actual threshold varies by operating system and storage system, 256 snapshots is typical.
What level of application integration exists?
Whether using CDP or near-CDP, when protecting applications like Microsoft Exchange, Microsoft SQL Server, Microsoft SharePoint or Oracle, it needs to integrate with the application in order to support transaction consistent recoveries. However, when using CDP, backup administrators can often fully recover applications. Application or database administrators may need to replay transaction logs to fully recover the application with near-CDP products.
Can your existing data protection software manage the CDP or near-CDP software?
A distinct advantage for near-CDP products is that most include Microsoft Volume Shadow Copy Service (VSS) providers. This allows data protection software to detect, schedule and manage snapshots. CDP software is often managed apart from data protection software. However, more data protection products are integrating with in-house and third-party CDP products.
Do dependencies exist between multiple application servers that create different, specific recovery points for each application server?
It's common for companies to have application servers that rely on data from other application servers. If data is corrupted or a failure occurs at some point in this chain, it can have a domino effect. The corrupted data spreads to other applications that affect them at different times. This requires administrators to recover data to different points in time for each application. Companies looking to recover from these types of disasters without data loss should use CDP software because it provides many more recovery options than what near-CDP can provide.
About the author: Jerome M. Wendt is lead analyst and president of DCIG Inc. You can check out his blog here.
