If i were to organise data on tourism and hotel for the continent of Australia and south America can you let me know a best way to do it. Like if i organise the tables thus:
It will be search oriented with countries, states, zipcode, cities and then amenities or features of hotels and tourist destinations
how would a hotel exist in multiple places? also, is this really one large table, or do you mean a large database?
i would suggest you model it using star schema. have a region dimension. would help in slicing and dicing of data.
In a nutshell - create one table for one entity.
I think you should create an ERD diagram (http://en.wikipedia.org/wiki/Entity-relationship_model), and map it to the tables in the following way: one entity - one table. As a result you will get a data in the 3-rd normal form. You should proceed this way until facing some performance problems. If they will be serious - start de normalizing. I do not think that building something not normalized is justified in this case.