Definitely. I agree with everything in this article. I am glad Erin writes this article as DBA best practice guide.
Alternatively, you could record troubled server by checking counters for CPU, memory, IO with system scheduled jobs while using triggers to record information for troubleshooting.
This is like when in trouble, patients automatic log symptoms, pains, temperature, color-picture and other diagnostic data to be used by doctors later. The principle works unless the server dropped dead. In that event, you will have the last black-box recorder data like before an airplane went down. That is why Oracle has DBA_History..... tables/views.