Google's John Mueller responded to a detailed technical question regarding ccTLDs, wildcard subdomains and the likelihood of it being seen as doorway pages in a Stack Exchange thread.
In short, the webmaster wants to create websites that would target city names around the world, whereby TLD for each countries will be used. Here is the mapping they were considering:
www.website.de for let's say Germany and www.website.cn for China and the sub-domains will be www.berlin.website.de and www.beijing.website.cn, as an example.
Google's John Mueller responded that this won't be a good idea.
Let me share John's full response since it can be a bit technical:
My first thought when reading the question was that this is going to be a case for the web-spam team. Please don't create tons of sites that are essentially doorway pages. Also, using wildcard subdomains (assuming the idea is to map them to cities after DNS resolution) make it extremely hard to determine how those URLs should be crawled.
Additionally, I think it's important to mention that a site using this kind of URL structure won't see any unnatural advantage in search. Search engines are just as good at handling URL parameters, there's no need to make it look like a website focused on [cityname],[countryname] when it's essentially just a part of the same website. Unless you have very good reasons to do this outside of web-search, I would recommend simplifying things significantly.
For geotargeting, using a ccTLD is a good way to let users & search engines know about your target audience. For Google, you can also use a gTLD (even the same one for all your sites) and work with subdirectories or subdomains to apply geotargeting there too. That saves you from having to get & maintain all those ccTLDs.
For multilingual content, at least from Google's point of view, the URL is irrelevant as long as it's unique. Use whatever URL structure works for you (and look into using hreflang where it makes sense).
Venturing overseas with content, especially with travel destination like sites, can be very tricky. This is especially true in the Panda world.
Forum discussion at Stack Exchange.
Image credit to BigStockPhoto for global servers