[geeks] faster www cache

[Charles Shannon Hendrix]

Thu, 07 Sep 2006 @ 14:43 -0500, Lionel Peterson said:

> >From: Charles Shannon Hendrix <shannon at widomaker.com>
> >Date: 2006/09/07 Thu PM 12:15:11 CDT
> >To: The Geeks List <geeks at sunhelp.org>
> >Subject: [geeks] faster www cache
> 
> >I've been running squid for years as my WWW cache.
> >
> >Yesterday I decided to look at the statistics and they are
> >disappointing.
> >
> >I'm the only user on my personal LAN, and I so I have a smallish 1GB
> >squid cache.
> >
> >I only get about an 8% hit rate, which is making me wonder if there is
> >any point to running it.  It doesn't even make a blip in process time so
> >I'm not worried about that part, but I wonder if I'm getting any real
> >benefit to using the caching.
> 
> You'd have to return to cached pages (or pages that refer to cached
> elements) before they expire to see any benefit from the cache. 

The pages should not be expiring that much, as I tend to hit the same
pages a great deal.

> Also, the pages (or elements) you revisit are most likely cached in
> your browser cache, preventing squid for "scoring a hit".

As it turns out, my browser disk cache was 50MB.  I set it to 1 MB
several times, but today I check and it has reset again.

Firefox and Mozilla are famous for resetting a few options for reasons
unknown.  Every single time I upgrade, the reset things like middle
mouse usage and other annoyances.  

If wanted that crap, I'd run Windows.

Anyway, the memory cache is no longer settable these days so I can't
turn that off, unless someone knows some hidden method.

> >Could it be squid just isn't appropriate for personal use?  Tuning
> >issue?
> 
> My gut says that unless you have a small browser cache, and you vist
> the same sites repeatedly, there is minimal benefit to running a cache
> for a single user. If you re-deployed that one gig as the browser
> cache I think you'd see a bigger benefit (when the browser cache is
> hit, the netowrk overhead is eliminated if retrieved from local
> cache)...

Well, here's the deal though.

You cannot control memory cache any more.  I don't see any way to even
know how effective it is or even what it caches.

about:cache is no longer an active URL, at least in the build of Firefox
I have.

So, I really don't see a way to fix that problem.

I also don't really want 1GB of browser cache in my .mozilla directory.

The other issue, and the reason for running squid, is that quite a few
different programs on my systems use the cache, so I figure a shared
cache would be better.

Some of the programs constantly hit the same pages over and over, so I
thought the hit rate would be decent.

It's also possible that the stats are distorted due to low hit rate in
the beginning.  The current cache is about 6 months old and around
900MB.

I don't know any way to reset the stats or otherwise figure out recent
performance.

Anyway, I will probably keep squid for a couple of reasons:

	- security, some programs can only get out through squid
	- shared cache, some programs might be getting a pretty good
	  hit rate since they hit the same thing all the time
	- it saves a bit on having multiple redundant caches here
	  and there

I really think it sucks how you can't ask a browser how its cache is
doing, or control the size of it.



-- 
shannon "AT" widomaker.com -- [There is a limit to how stupid people really
are -- just as there's a limit to the amount of hydrogen in the Universe. 
There's a lot, but there's a limit.  -- Dave C. Barber on a.f.c.  ]