We could use many examples of large datasets that could benefit from partitioning. In our case, we will use a contour dataset. Contours are useful ways to represent terrain data, as they are well established and thus commonly interpreted. Contours can also be used to compress terrain data into linear representations, thus allowing it to be shown in conjunction with other data easily.
The problem is, the storage of contour data can be quite expensive. Two-foot contours for a single US county can take 20 to 40 GB, and storing such data for a larger area such as a region or nation can become quite prohibitive from the standpoint of accessing the appropriate portion of the dataset in a performant way.
