In the previous section, you used an OpenLayers map of width 600 and height 400. The server sent you a single image of that size when you used the Single tile option. Switching to the tiled mode, OpenLayers split the map in tiles of 256x256 pixels. So, it sends 9 GetMap requests to GeoServer. In the following image, you can see the blue rectangle covering the 600x400 area of the map, while the 9 256x256 squares are the tiles:
Each request to the server is the same for any single tile, except the bbox parameter specifying the area.
Note
The bounding box parameter is called bbox. The value for bbox is the latitude and longitude of the area you're calling from GeoServer. The format for this parameter is bbox=minx,miny,maxx,maxy.
If your map's height and width are fairly small, using a single tile will likely take less time to render. This depends on your data filter and number of features too, but it is a good rule of thumb. Using a single tile will also be useful if you need...