In the previous recipes, you used a single raster dataset. This is a common case when the datasets are not of a huge size. When your data covers a really big area or its resolution is high, that is, it contains a highly detailed representation of the measured parameter, a single dataset is not the best option, and in some cases is not an option at all.

The best approach is to create a mosaic of raster datasets. A mosaic is a collection of raster datasets that are virtually merged together to be used as a single dataset.

The virtual merging operation produces an index file that contains information on the extent of each raster dataset. Using this index file, GeoServer compares the area of each request and retrieves the granules of the mosaic that intersects it.

You may build this index file, which is actually a shapefile, with an external tool. GeoServer has a built-in plugin to create a new mosaic, which is the index file, and it works really well, making it the best option...