Since IGD is implemented with ACeDB, its size is limited by ACeDB. Currently, a single ACeDB database of IGD's complexity cannot handle much more than a few gigabyte of data (aprox. 1 million objects). Some of these limitations can be overcome by partitioning the data and maintaining it in a parallel multi ACeDB system. We are not aware of the scalability of Docking-D (i.e., its underlying VODAK DBMS).
Entrez, SRS, and LinkDB do not seem to have scaling problems.