Oracle Performance Tuning Blunder #11

Just Add more Memory!

The “increase the parameters” strategy has a certain appeal, because it sounds so logical and scientific. In this line of reasoning, performance problems will disappear as the cache hit ratio increases. 

At first, this sounds logical since a high cache hit ratio means fewer disk reads. The grand fallacy with the “increase the buffer” strategy is that it fails to address the root cause of the problem. 

In the rush to achieve a solution, no time is spent calmly analyzing the facts and finding a root cause that matches the facts.  That is, just like a “guessing” strategy, a sound process is not even considered as the analyst hopes to find a solution that doesn’t require any real work. 

The approach of "add more resources"  is nothing more than guessing. The problem with guessing is that there are unlimited number of guesses one can make. If you ask around the office, everyone can make a guess, no matter their competence in the subject matter.

Of course, it won’t do any harm to increase the database cache size, but it’s a big distraction.  Whenever anyone suggests this approach, I ask, “So, that’s the root cause of the problem?”  This question is sort of a polite way of asking, "Do you really know what you are talking about?"

In my entire career, I have not seen a single case where the root cause was actually insufficient database cache.  (However, a colleague told me that he saw it once.)