The thought just occurred to me that disk thrashing is mainly
due to SEEK time, so we should cache metadata more agressively.
And now that we have quota for both buffer and page memory, it
would seem best to do page-aging on buffer cache as well, since
we now just discard a random buffer page, while we could have
chosen one more specifically...
Of course, this brings overhead and all new exiting trouble,
so first I'll ask if it's worth it...
And: Bandwidth is all, or is it?
Rik.
+-------------------------------------------+--------------------------+
| Linux: - LinuxHQ MM-patches page | Scouting webmaster |
| - kswapd ask-him & complain-to guy | Vries cubscout leader |
| http://www.fys.ruu.nl/~riel/ | <H.H.vanRiel@fys.ruu.nl> |
+-------------------------------------------+--------------------------+
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu