The results of a performance comparison between relational MySQL and object-based database for objects (DB4o) database systems are presented in this paper. The comparison was restricted to insert, select, and delete operations. The data used was flight information characterized by 29 attributes. For insert operations, DB4o was found best. Otherwise, MySQL was found best.

Unfortunately, much criticism can be made of this research. Other than naming the two database systems used, the technology platforms are otherwise not described. There is no discussion of benchmark approaches. In particular, there is no discussion of an earlier, more comprehensive study, which made use of the PolePosition benchmark to compare DB4o with relational databases [1]. There is no discussion or use of indexes, which can dramatically impact performance times. The authors describe having used import mechanisms for the insertions, yet provide MySQL and DB4o code examples for the insertion of a single record only. It is unclear what kind of bulk loading operation was used when inserting thousands of records. The authors describe using strings for all 29 attributes, yet no real database would encode data like this.

The performance graphs do not explicitly indicate the units of time. The graphs and tables suggest only single trials were run, making it important to know how accurately time was measured. Unfortunately, information about how time was measured and the timing accuracy achieved is not provided.

Many researchers will regard this study as weakly conducted and poorly reported. This paper will be of interest only to those conducting performance comparisons of database systems.