[bobfunk]

I've done lots of these kinda performance tests against all kind of dynamic sites, and higher end managed hosting services like WPEngine, etc...

Once in a while someone manage to get CDN hosting just right, but it's really rare, and it's not something you can simply automate with a dynamic site (like we can for static sites with netlify). Typically the result is identical to the Smashing Magazine Site, often a lot worse. Smashing does a good job of caching at their origin datacenter, but their HTML doesn't get cached at edge nodes. Many other sites does a far worse job of caching at their origin.

It might be true that to some degree it's an engineering issue, but if it's one that hits 95%+ of all sites built with a dynamic approach and can be completely eliminated with a static approach, then obviously it might be better to shift the balance and default to doing thing statically instead of reaching for Wordpress/Rails/Drupal/whatever for each new site...

[z3t4]

I think CDN is over-kill/hype. If you do everything right, all you get is better latency.

If your dynamic site loads slower then a static site, you are probably doing needless database round-trips, redirects, synchronized writes, or html rendering.

[ChristianBach]

Yeah, caching HTML directly on a CDN basically only gives you better latency.

..Which in turn gives you better page rank. ..Which gives you more traffic.

But that's it.

..Well besides that it also gives you lower bounce-rate. ..Which means higher conversion. ..Which means higher ROI.

So there's that.

:-)

[z3t4]

That is best case scenario ... But probably a premature optimization. You also have to look at time to first byte, total time, and client rendering time.

If it takes like ten seconds to render the site on the client, then 10ms gain on connection time wont help much.

Test tool: http://www.webpagetest.org/video/

When all css, fonts etc are cached on the browser client, then there's almost no gain with CDN.