Yee Man Chan wrote:
> ... I want to do pure LRU, pure LFU, pure SIZE, etc ..
Just curious, why would you want to do that?
There are already quite a few papers out there that study and compare all
these algorithms. Also, for real (i.e. large) caches all good algorithms give
about the same hit ratios. There is not much one can optimize in terms of HRs
when "everything" is already cached.
Think about a large, Pure-LRU-TH-driven cache. Most purged documents are,
say, 4-5 days old. Thus, you are not choosing victims among "good" and "very
good" documents, but only among "very bad" ones.. Any reasonable algorithm
will perform about the same because it does not matter which "very bad"
document to purge.
Most of the algorithms you have mentioned (and many others) came from
Databases and OSes where caches are relatively small and "hot".
$0.02,
Alex.
Received on Thu Mar 05 1998 - 00:00:44 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:39:09 MST