Hello WebmasterWorld!
Hope this is an appropriate for a first post...
In attempt to bring the best experience to users while shooting for cost efficiency, I have split up content into two general parts for media rich websites. The first part includes dynamic pages, databases, web applications, stuff that changes, etc, into the cloud. The second part is all the static content such as images, video, flash, static files, etc.
With a priority on serving users a similar experience globally, I am about to implement a CDN solution for the static content. There appears to be two options I have with this implementation.
For a site such as "example.com" that uses cookies to maintain state:
1) Configure a CNAME such as "static.example.com" for the CDN.
This will achieve parallel downloads, utilize urls matching the primary domain, but serve wasteful cookies with each object.
2) Configure a CNAME such as "www.example2.com" for the CDN.
This will achieve parallel downloads, serve content cookieless, but utilize urls unrelated to the primary domain.
The question I have is how does Google assign attribution for content on a CDN? I'm concerned that by implementing such a solution, Google will mistake original, media rich sites as scraper sites, thus affecting future rankings and traffic. I believe that goes for either option as anyone can CNAME a domain to a CDN url (when the CDN company permits it)?
At least with the Page Speed recommendation, the CNAME of the cookieless domain is tied to the A record of the primary domain.
I've spent a few suns and moons researching internet on the topic with no definitive answer. This includes searching for a web server hack, setting up a web app cookieless, and a few other angles I cannot immediately recall. I have thoughts for setting up a subdomain alias, then writing some kind of web server rewrite hack to get the cookie to operate in the primary domain by tricking the web app, but I haven't gone that far yet. And the only metric I can think of that would tie a new domain to CDN urls in this context would be a domain's registration details, which could be faked for such purposes as well.
Thanks to all who see this. Insights into which way to go, or just a good discussion of the topic is greatly appreciated!
[edited by: tedster at 5:03 am (utc) on Feb 7, 2010]
[edit reason] switch to example.com - it can never be owned [/edit]
