Cost of too much database access

I'm wondering if my idea is possible*. I'm doing a project on road networks and I plan to do many database access; meaning, for exampl开发者_如何转开发e, to check if one route is connected to another route, check database if connection exist? I have predefined information about the routes so I want to use them as mush as possible, but I'm worried that too mush database access will make the program slow. I want to know how Google does it in their Get Direction feature of the Google Maps, but I can't find the right article that describes the method directly. Please help me. Thanks a lot!

I'd first take a look at:

http://en.wikipedia.org/wiki/Breadth-first_search

vs

http://en.wikipedia.org/wiki/Depth-first_search

And then progress to:

http://en.wikipedia.org/wiki/Minimum_spanning_tree