Hi, just finished reading this excellent article giving great experimental data to explain why looking simply at buffer cache hit ratio isn't good enough.
https://www.simple-talk.com/sql/data...che-hit-ratio/

However, the reference to "Page Life Expectancy" is I believe SQL Server specific. Heck I can't even find any reference to it with Sybase despite the two are really the same product originally.

Does anyone know what is the equivalent of PLE in the Oracle world?

The example also provides some figures for disk IO tying its relevance to how it may be the bottleneck despite us observing a high comfortable buffer cache hit ratio figure, because we may before we realize run into a limit of disk IO rate. The high buffer cache hit ratio may be due to read ahead mechanism being used. So the high figure is due to lots of page churn so to speak. This obviously is benefited by doing the check locally on the box. Just wondering what other type of performance checks we won't be able to do if this is done remotely (ie no OS level access and only database access). Any ideas?