Best Practices: Foolproof DR is still a moving target
This article can also be found in the Premium Editorial Download "Storage magazine: Multiprotocol arrays provide NAS and SAN in a single box."
Download it now to read this article plus other related content.
- Establish mature operational disciplines. One of my colleagues is fond of pointing out that one of the best ways to improve DR is to improve production. Put another way, if normal day-to-day operations don't tend to function well, your DR isn't likely to either. Therefore, operational discipline is an essential element of predictable DR. The first sign of a potential operational deficiency is the lack of documentation for key processes. Given that DR, by definition, occurs under seriously sub-optimal conditions, the need for well-documented standard operating procedures is clear. Organizations that have established and actively embrace standard frameworks, like the Information Technology Infrastructure Library (ITIL), are significantly improving their odds of recoverability in the chaotic atmosphere of a disaster situation.
- Develop a realistic testing methodology. Given the operational disruption, practical difficulties and costs involved, we tend to focus our testing on those components that are easy to test. But realistic testing is just that--testing real business function recovery. While it's necessary to perform component testing on a regular basis, it's equally important to test the recoverability of large-scale functions to ensure that interoperability and interdependency issues are addressed. The closer
- to a real production environment a test can get, the more "provable" the DR capability.
The elements outlined here transcend the boundaries of the IT infrastructure. It's therefore critical for IT administrators to have a strong understanding of the problems at hand and to learn how to address them so they can influence strategic decision-making wherever possible. This will help them avoid being placed in the Catch-22 situation of solving a problem over which they have no control.
This was first published in March 2008