Exo Research CTO still does not understand how OS file caching works in Windows

Ok, I made the comment in <a href="http://news.ycombinator.com/item?id=1138840" rel="nofollow">http://news.ycombinator.com/item?id=1138840</a>, that "it's surprising that the CTO of a company specialized on system performance doesn't know how OS cache memory works." It turns out that he really has no clue.<p>To the people who said the Windows source code is needed to understand how it works, hogwash. The file caching in Linux has been doing it this way for years. Windows does it for years as well. It's fairly standard OS caching technique.

Actually, there was one nugget of truth in the article, and that was the _best_ way to determine if your operating system is low on memory, is to watch it's swapping activity. This is probably true across all operating systems, Linux, Solaris, etc..<p>An operating system will typically not swap out (or in) cache memory to disk, so, as long as you are exhibiting no swapping behavior, (And you have swap enabled) then you have enough memory. I think it was Adrian Cockroft, in fact, in "Sun Performance and Tuning: Sparc &#38; Solaris" who in fact said that was _the_ way to determine if you have memory pressure on your operating system - run vmstat and watch the si and so column for activity.

Heh. I've been reading the various discussions about all of this, and having considerable experience digging into internals to find performance problems, it still seems to come down to one thing.<p>Does it _feel_ fast?

Regardless of whether the CTO is correct or not, the tone and poor attitude of his posts have convinced me that he is not someone I would want to do business with.

I'm not even sure the facts in this case matter if the CTO is willing to make such a fool of himself in public. Even if I agreed with his point of view, it's impossible for me to take him seriously when he writes such needlessly incendiary rants on his company's official blog. It reminds me of conspiracy theorists.