Automation reduces errors

I recently wrote about how automation can improve efficiency, reduce downtime and free IT staffs from mundane tasks so they can do more “important” work. One of my readers felt I had left out an elemental benefit of automation: reducing errors. He’s right that there are many more aspects to automation than I have discussed, so today’s column will discuss more of these aspects.

Reducing errors is an important result of automation. Many problem-resolution actions that operators or administrators perform are often conducted under the stress of having to remedy a situation. In some cases, users may be breathing down their necks, wanting a quick resolution. This kind of environment opens the door to errors.

One of our readers reports that “research (by a major vendor) into extended outages showed that 60% of them could have been significantly shortened if the operators had followed documented procedures. This console automation would have trapped the reported error and then followed predefined recovery procedures.” Once automated responses have been thoroughly tested and used on a regular basis, they could minimize problems caused by operator error.

Management tools can also help to reduce errors in configurations. Another major cause of problems is changing configurations. As systems and network equipment is deployed and reconfigured, with all of the options and features that must be configured, configuration errors are likely. In some cases, these changes may propagate more downtime or reduce performance. Taking proven configurations and automatically moving them to new devices minimizes the effects of configuration errors. Managing your configurations so that they are consistent across the infrastructure can also help.

As I discussed in my previous columns on automation, many vendors are setting their sights on automation. Whether it’s called autonomic, adaptive, or whatever, the benefactor of such initiatives will be you, the IT staff. Not only can you benefit from the automation these tools will provide, but there is also another underlying benefit that is not as evident: integration. No longer can each tool manage its own silo well – the tools must be integrated and intertwined before true dynamic autonomics can be achieved.

That’s in the future, but interesting to ponder nevertheless. For example, if performance is slowing down due to a lack of compute power, a performance management tool must be able to communicate that event through an event management system. The event management system must then connect to a provisioning management system that can quickly provision additional computing resources and bring it on line. The provisioning system must then report back to other systems that the provisioning is completed. And the performance management tool must monitor the effects of the change, and make adjustments based on the results.

So if you think integration of management tools is an important issue, as we move into the “on-demand” world, integration will be essential.