Payment protocols for electronic commerce have traditionally been studied mostly by the cryptography, security, and distributed computing communities. However, as pointed out by this invited paper, the database-style notion of atomicity is of crucial importance for such protocols as well and has been neglected so far. In fact, one could argue that keeping distributed data consistent in the presence of server failures is an absolutely critical concern in electronic commerce, whereas perfect protection against tampering, albeit highly desirable, is not the most pressing issue given that a decent suite of security measures is already in place. In fact, being billed for some goods that one has never received or accidentally receiving some ordered goods twice may turn out to be as troublesome as a stolen credit card number.
Using atomic transactions for the distributed processing across multiple servers on behalf of the merchant, the bank, and the customer is, of course, state of the art in the database community. Regardless of the fact that many deployed e-commerce solutions may exhibit severe engineering deficiencies in this regard, adding distributed transactions does by itself not pose any research challenges. Rather, it is the combination of payment protocols and transactional protocols that needs to be studied carefully. For example, to what extent do protocols for anonymous payment and atomic commit influence each other?
This paper provides an excellent overview of the issues in this contemporary area where payment and transactional protocols need to be reconciled. It points out a variety of open problems and research opportunities in the intersection of these two avenues (beyond merely emphasizing the need for atomic transactions in electronic commerce). I highly recommend reading this paper and particularly its Section 7 on open problems to everybody doing research on electronic commerce.
Copyright © 1999 by the author(s). Review published with permission.