Are you running the server locally or connecting to it over the internet? I think it's really only meant for local use, and if you find it too slow for some reason then you can just use the single file version in the regular way. Having all the tiddlers as individual single files is incredibly useful imho.
If you're trying to use it across the internet, I'm afraid I don't think it's secure and possibly not too stable (I played around running it on a digital ocean droplet (hosted VM) and connecting to it and found that the droplet fell over quite a lot and performance was generally dire).
Another backend that was written is called TiddlyWeb and that still powers TiddlySpace. A developer called Chris Dent has been working on another backend called Tank.
Also, Danielo Rodríguez has been working on a couchdb adaptor, which would allow you to use a database on a remote machine as your tiddler store
https://github.com/wshallum/couchadaptor - and various other methods like this using other technologies are also possible if someone writes the code.
As far as I know, there's no way to pass url strings to the search box - it puts the list of fragment (tiddler) identifiers into the story list. If you enter the names of tiddlers that don't exist (
http://tiddlywiki.com/#newtiddler) it will give you a 'create tiddler' dialogue for each one.
I wonder if Jeremy has ever considered a more sophisticated 'routes' mechanism? Probably a massive can of worms!
Regards,
Richard