I am playing around with the Google Maps JS API, Mapbox API and I was curious, how are markers actually added to a map? Does the backend code take a set of map tiles, convert the edges to lat/long, and then simply interpolate the lat/long of the marker coordinates along the X and Y axes?
I ask because I am building an application that would need 1000-5000 simultaneous markers, and want to build my own backend system as using the Google Fusion Tables API can get expensive.
how are markers actually added to a map
In the simplest implementation, markers are represented as GeoJSON or similar, and requested from the server, and then 'projected' into screen coordinates from lat/lon; see node-sphericalmercator for one example of that logic.
For the 'performant' case, like TileMill or Google's pre-rendering mode, markers are baked into raster tiles, PNG files, that show them and their locations, and you do magic like UTFGrid to do interaction.
Note that both of these are compromises: you can get speed, flexibility, and simplicity, but it's very hard to get all at the same time.
