a question about Andres's paper on LRU

From: <maer727@dont-contact.us>
Date: Mon, 8 Apr 2002 15:31:22 +0800 (CST)

Hi, pals!

I have read the paper written by Andres on LRU from
http://www.squid-cache.org/Devel/papers/kroonmaa-refresh-proposal.txt. It is a nice material
to integrate LRU and Squid together.

I am also interestied in LRU and Squid. I have two questions about the paper.

1. In the paper Andres mentioned,
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
   Estimate penalty of fetch - size, time needed, delay, hopcount to source,
   estimate cost of retrieving - pattern configurable billing weight,
   possibly taking into account hopcount?

      penalty = size*SizeFactor + time*TimeFactor [+ hopcount*HopFactor]
      cost = penalty * PatternFactor,
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
I do not know what means "pattern configurable billing weight". I also do not understand the
meaning of "PatternFactor". What is PatternFactor idebtied for?

2. In the paper, Andres mentioned,
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
   Pure hit increases average hitrate and hit count. Object has been once
   more time useful to be kept, it has conserved bandwidth, time and thus
   some money.

      refcount++;
      usefulcount++;
      ( and, bytesaved = size*usefulcount )
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
What is the function of the variable "usefulcount" here? I noticed that in the latter case,

/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
   Refresh or reload decrease or reset object's usefullness and implicitly
   reduces average hitrate and conserved money.

      refcount++;
      usefulcount -= somefuncof(cost);
/////////////////////////////////////////////////////////////////////////////////////////////////////////////////
refcount increse, but at the same time, "usefulcount" decrease. So I am puzzled at the
meaning of the variable "usefulcount" here.

Can anyone give me a brief explanation?

Best regards,
George, Ma
Received on Mon Apr 08 2002 - 01:31:26 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:58 MST