I am evaluating a database for my next project. I wish to store all of the metropolitan areas on the planet (2,5 million) and save weather forecast for each city every single day. So that you can think the dataset can get quite large fast.

I should also perform geo queries - get me the town and temperature with this day within this bounding box.

To date I have checked out hbase and couchdb. Hbase looked interesting, however the hardware requirement of production is simply too costly for me personally (an exhibition stated you'll need 5 separate servers).

Let me keep your costs to a minimum, it's my own project.

What exactly other available choices have i got? Can mongo handle this ammount of information? Other things?

TLDR The needs are

  • Great deal of data
  • Fast bounding box queries
  • Low/cheap hardware needs
  • Enhanced for read, but must handle place of two,5 million records daily

Yeah, you are able to opt for mongodb. Mongodb was created for scaling (sharding, replication). Additionally mongodb support geospacial search.

Strange question since MongoDB is marketed as large-scale database...please develop more specific needs. MongoDB has enough choices to partition data eg. through Sharding.