Re: [SQU] lru or heap LRU (Squid V2.4)

I don't think the HEAP implementation uses a high/low watermark. If it
does then it should be the same as for Squids standard LRU policy.

The main difference between the two policies (apart from implementation)
is that Squids standard LRU does tail compression by removing expired
objects from the tail. This causes some objects at the tail to be
scanned many times.

Technically the watermarks in Squid is not watermarks any longer. There
is a ramp function in that the closer to the high watermark the usage is
the more aggressively Squid expires objects.

--
Henrik Nordstrom
Squid hacker
Randeep Cheema wrote:
> 
> Hi Val,
> 
> Studies & simulations shows that LRU performs little bit better tha LRU_H
> (LRU Heap) in terms of number of objects served from squid-Cache. But LRU
> tends to be more CPU intensive than LRU_H. Now it depends on your
> requirement that whether you want to serve more objects from cache or you
> have slower CPU at proxy.
> 
> Henrik,
> 
> Could you please tell me what are the values of watermark in Squid
> implementation for LRU_H.
> 
> Thanks
> ~ Randeep
> 
> -----Original Message-----
> From: Valentin Chopov [mailto:valentin@valcho.net]
> Sent: Sunday, October 01, 2000 8:43 AM
> To: Henrik Nordstrom
> Cc: squid-users@ircache.net
> Subject: Re: [SQU] lru or heap LRU (Squid V2.4)
> 
> Henrik,
> 
> thank you for the quick reply. Do you have any info about which of both
> impllementations gives better results?
> 
> Thanks again,
> 
> Val
> 
> On Sun, 1 Oct 2000, Henrik Nordstrom wrote:
> 
> > Valentin Chopov wrote:
> > >
> > > Is there any differencies between  both cache replacement policies?
> >
> > The two policies are similar, the actual implementation is very
> > different.
> >
> > HEAP LRU is provided as a reference to compare the HEAP implementation
> > to the "normal" LRU policy.
> >
> > --
> > Henrik Nordstrom
> > Squid hacker
> >
> > --
> > To unsubscribe, see http://www.squid-cache.org/mailing-lists.html
> >
> >
> 
> --
> To unsubscribe, see http://www.squid-cache.org/mailing-lists.html
--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html