On Mon, 21 Feb 2000, Nikos Chrysos wrote:
> does anyone know if squid how squid decides when to start
> replacement...? is it based on the number of objects , or on the
> actual occupied size
The short answer is the portion of the defined cache_dirs that are
occupied - Squid can handle (almost) any number of objects). The long
answer is that Squid uses a number of factors to determine when to discard
objects: if an object reaches 'reference_age' then it is discarded; when a
cache_dir reaches the low-water mark ('cache_swap_lo') the least recently
used objects are discarded at such a rate as to attempt to maintain the
cache_dir size at the low-water mark (the aggression of the discarding
increases as the cache_dir size approaches the high-water mark).
To fine tune your cache, different replacement algorithms are now
available: LFUDA and GSDF - read squid.conf for details as to why you
might want to choose these, and then read the Hewlett-Packard paper on
their different performances under test conditions (I can't remember the
reference, but it's around somewhere - try www.squid-cache.org if you're
interested).
-- Kendall Lister, Systems Operator for Charon I.S. - kendall@charon.net.au Charon Information Services - Friendly, Cheap Melbourne ISP: 9589 7781Received on Sun Feb 20 2000 - 22:35:37 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:51:22 MST