I am finishing the Squid implementation of the alternative
replacement policies I presented at WCW this year. At present I have
implemented the policies within #ifdef HPL_REPL so people can compile it
in or out according to their preference. Now I am creating a config
file option to control which policy is selected (rather than determining
this at compile time).
I am using the IFDEF tag in cf.data to control whether the
comment about the policies is in the file; this is cool. However, I do
not see a way to generate an #else for the #if, or an #ifndef test for
HPL_REPL. I need this because the referenceAge is no longer meaningful
when compiling using one of the non-LRU replacement policies. I could
just leave it in with a comment, but it would be cleaner to remove it.
Please suggest a way to handle this case. We currently have the
code running and tested under a synthetic workload, as well as on some
internal production caches. So far it's performed well and I plan to
submit a diff against 2.2.STABLE3 shortly. Thanks for your suggestions,
-- jad --
John Dilley <jad@hpl.hp.com>
PS.
I have updated the WIP report I presented in March at WCW with
current performance results (and after resolving the anomaly that was in
the results in my presentation). You can find a copy of the report at
http://www.hpl.hp.com/personal/John_Dilley/caching/wcw.html and a full
description of the replacement policies at
http://www.hpl.hp.com/techreports/98/HPL-98-173.html.
Received on Tue Jul 29 2003 - 13:15:58 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:09 MST