🚀 go-pugleaf

[POSTED TO comp.dcom.modems.cable; PLEASE REPLY THERE]

In <3A8D6226.C953CFEA@mynetwatchman.com>, Lawrence Baldwin
<baldwinlSPAMMENOT@mynetwatchman.com> wrote:

>OK, I made some suggestions that cached connections can be slower (in
>some cases) than non-cached.
>
>Let me emphasize that I am NOT saying that caching is bad...in fact I
>believe a well engineered caching architecture is essential to providing
>consistent performance at a reasonable cost, so I'm all for it.
>
>I am merely interested in understanding all the technical details and
>variables that come into play when you introduce caching.  So here's
>some hard data, including packet traces for those that are interested:
>
>http://www.mynetwatchman.com/kb/adsl/
>
>(See Case Studies: Cache Comparison)

   Observations:

   ...
   Non-cached connection reused the same TCP connection for each page
      load 
   Cached created a new TCP connection each time the object was
      requested, and tore it down as soon as the object was loaded 

That sounds to me like you retried the cached access without restarting
the browser or at least going someplace else.  I don't think that models
the real world terribly well, biasing the results against caching --
most users do keep pounding on the same page.  OTOH, access on just the
first trial was markedly faster for cached versus non-cached.  Sample
size issues aside, I'd say that cached access was actually the clear
winner (as I would expect).

-- 
Best regards,
John Navas     <http://navasgrp.home.att.net/>
     CABLE MODEM/DSL GUIDE:  <http://Cable-DSL.home.att.net/>