🚀 go-pugleaf

John Navas wrote:
> 
> With all due respect, that's oversimplified and based on assumptions
> that aren't necessarily valid, depending on the particular proxy and its
> configuration.  I've never seen a case where a properly configured
> Cacheflow of appropriate size was slower than a non-cached connection.

John, I'm talking about service provider proxy cache servers (located in
their core network), not individual/private cache servers, such as I
assume is your configuration.  

And absolutely, there are a lot of variables, but I thought I did a
pretty good job of sorting out *some* of the variables:

> >Here are cases where cache servers will likely improve your experience:
> >
> >* You frequently access popular content
> >* Cache server "close" to you and high-powered
> >* Providers backbone connections are congested
> >* Real servers are slow
> >
> >Here are cases where cache servers will likely degrade your experience:
> >
> >* Accessing obscure or highly dynamic content
> >* Cache server is "far" away and/or underpowered
> >* Providers backbone is NOT congested
> >* Real servers are very fast (e.g. large web farm with local
> >load-balancing/caching)

My whole point was that there are just so many variables, that there is
no easy way to determine (without protocol analysis) if they are helping
or hurting you.  Providers like subscribers to assume that
"cache=good"...my other objective was to demonstrate that this is not
unilaterally true.

I wonder how many providers actually tested user experience (using
analyzers) before deploying a caching strategy.  I think many also
assumed that it can only help, not hurt you.

lb.