Question

Hi I’m writing an Android lbs game, I need to:

• store user’s real time position(lat, lon)

• get users that are near to a given position

I’ve wrote some back-end api with django and mysql(for example, get user name and age by his id).

what’s the best solution to do what I said above (with django/python)?
Should I use a nosql database like redis?
And how should the positions be stored in database so I can quickly get positions that is near to a given position?

Answer 1

Well, I am a enthusiast of Redis, but considering your specific problem, there is a NoSQL database engine that better supports geospatial queries: MongoDB.

I have never used MongoDB, but in conferences I’ve seen people talking about this feature: http://www.mongodb.org/display/DOCS/Geospatial+Indexing

I read through it carefully now, and focused on these main points:

• The geospatial index has to be mantained for every insert in the database (i.e. generated again), which differs from Redis, where there is no such thing as an index (sorted sets in Redis “could” solve your problem, but again it would be a lot more work);
• You will use find and/or geoNear queries in MongoDB;
• For combined querying with other indices, you have to specify and combine all indices at creation time;
• Distance can be measured considering Earth curvature in geoNear (with the spherical attribute). If this attribute is not used (or find is), the distance is converted from a fixed value of miles per arcdegree.

Also a curiosity: the “logo” in the Stack overflow NoSql tag is from MongoDB.