In a typical multilevel cache (e.g. the ones we see on
microprocessors), data that's fetched from a lower level cache is
kept in higher levels of cache in case it's used again. Due to the
principle of locality of reference, this practice usually pays off.
However, in my use of Squid I haven't seen TCP_MEM_HITs when an
object that's in the cache is accessed multiple times, even in
rapid succession.
Does Squid "promote" objects from disk to RAM? And if not, why not?
If it doesn't, restarting a Squid cache could cause a substantial
loss of performance, because objects which were previously in RAM
won't be restored to it and will be fetched more slowly thereafter.
--Brett Glass
Received on Thu Jun 30 2005 - 11:14:34 MDT
This archive was generated by hypermail pre-2.1.9 : Fri Jul 01 2005 - 12:00:03 MDT