Besides Turing-completeness of Prolog and non-Turing-completeness of vanilla Datalog (there are extensions of Datalog that extends its complexity class), a couple of significant differences between Datalog and Prolog are as follows:
In Datalog, the order of rules, or atoms within the rules, do not make a difference in their evaluation or the results of the evaluation. In Prolog, on the other hand, order of atoms may mean the difference between termination and non-termination, or the runtime of a terminating evaluation. Prolog also offers imperative operators such as cut (!); typical Datalog implementations do not have such operators. Thus, Datalog is more declarative in that sense.
Datalog implementations typically supports bottom-up evaluation, whereas Prolog implementations typically supports top-down evaluation. There are advantages and disadvantages of each depending on how the program intends to be used, and the data size the program is meant to operate over.
Not sure the implication for your specific use case. It depends a lot on exactly what you're trying to build -- whether you need the Turing-completeness, etc.