In our article on disaster recovery planning, we looked at how to choose the right business continuity (BC) and disaster recovery (DR) KPIs. Now let's see how to optimize those KPIs so they meet your business goals.

Metrics are how we define success and failure, and different metrics call for different action plans.

This first part covers the recovery point objective (RPO) and how to optimize it.

1. Back up often

How often do you back up your data? With daily backups, your maximum RPO is 24 hours: you can lose up to 24 hours of data. The simplest way to improve this is to back up more often. Just keep the frequency reasonable.

2. Use replication and re-synchronization technologies

Copying data over the network lets you keep several up-to-date copies of your critical data. This brings your RPO down even further without having to back up unreasonably often. Put replication and re-synchronization technologies to work, and configure them to match your target KPI value.

3. Multiplex your critical files

Every system has files that matter more than the rest. In Oracle Database, for example, these are the control files and redo logs, and losing them seriously compromises your ability to recover the system. The answer is multiplexing: keep several physical copies of such files, so that if one is corrupted or lost, another is still there. You can do this in several ways, from built-in features (as in Oracle) to RAID arrays.

4. Take advantage of cold standby systems

Consider keeping one system as the backup for another, identical primary system. Unlike a conventional cluster or standby, a cold standby runs no services, so you avoid paying for compute, memory and licenses you are not using. Just make sure you can allocate all of those resources when the standby is promoted. This approach helps you optimize both on-premises and cloud DR solutions, and keeps license costs to a minimum.

In the next two parts of this series, we look at optimizing the other KPIs: first the recovery time objective (RTO), then the recovery window and system redundancy.