In this paper we investigate techniques for decreasing the overhead of semantic integrity enforcement or equivalently the overhead of transaction validation with respect to a set of semantic integrity (SI) assertions. We discuss the problem of semantic integrity enforcement from two points of view. First we describe three approaches to decrease the overhead of SI enforcement. Second we analyze the cost of several SI enforcement methods in centralized and distributed database systems based on slow and fast (local) networks.

--Author’s Abstract

I’m afraid I didn’t find this paper very valuable in assessing semantic integrity enforcement techniques. The three approaches suggested to decrease overhead seem terribly obvious, and their presentation seems somewhat superficial. For example, one suggestion is to associate with an integrity assertion X a condition Y describing circumstances under which a transaction could not violate assertion X. The paper does not point out that whether or not this actually decreases overhead depends on the complexity of evaluating condition Y, nor does it cite any examples of X and Y with cost comparisons.

It would also have been helpful to see in the presentation of the cost analysis some additional material to help the reader evaluate the validity of the conclusions, or even the validity of the simulation model used in the cost analysis.