What gain attention to me was two things: (a) On IBM page was suggestion to set this setting for MySQL database and (b) is the sentence from Linux kernel developers debate: "Whenever you read from a file, even one in memory cache.... do a write!" So if I understand correctly even if data in database are in bufferpool every time that data are accessed from memory an update of timestamp for atime is given. If I understand correctly this could have an impact. And then few posts later: "I didnt mean to say that it is _always_ a big issue, but "only a small number of files are read" is a very, very small minority of even the database server world."

You probably read this from Linux Tuning Parameters web site. And yes, we are using external storage, Linux on "System z" we are using Enterprise Storage Server.

We also don't have a IO performance problem. But what I am wondering is: "Do we have no disk IO problems because we have a 'big machine' and if we can lower down disk IO, could we have a smaller machine?"