How Google Maps works

Joel Webber has blogged an excellent analysis of how the astonishing Google Maps service is accomplished. Great place to start if you're trying to figure out how to build stylish, cross-platform dynamic Web apps:

The top and side bars are (more or less) simply HTML. The center pane with the map, however, is a different beast. First, let's address the map itself. It is broken up into a grid of 128×128 images (basically like an old tile-based scrolling console game). The dragging code is nothing new, but the cool trick here is that each of these images is absolutely positioned — and the 'infinite' scrolling effect is achieved by picking up tiles that are off-screen on one end and placing them down on the other end. The effect is kind of like laying track for a train by picking up track from behind it.

The push-pins and info-popups are a different matter. Simply placing them is no big trick; an absolutely-positioned transparent GIF does the trick nicely. The shadows, however, are a different matter. They are PNGs with 8-bit alpha channels. Personally, I didn't even realize you could depend upon the browser to render these correctly, but apparently (at least with IE6 and Mozilla), you can. And they actually render pretty quickly — for proof, check out the overlaid route image (at the end of the article), which is often as big as the entire map view.

Link

(via JWZ)