Online reorganization of databases

Reviewer: Ned Chapin

This survey of literature on system maintenance of large databases focuses on databases that are used 24 hours a day, seven days a week. Since such a database is continuously online, system maintenance must also be done online, on the fly, without taking the database out of service, even briefly. This extensive survey relies on a reference list that consists of 373 items. Sockut and Iyer assume that the reader will recognize that a database can be regarded as having a logical structure-how the accessing software views the database-and a physical structure-how the data is represented in the storage media where the database resides, such as a hard disk. They use the term "reorganization" to refer to official intentional changes in one or both of these structures. The authors do not address the effects of malware attacks or recovery from them, and they say little about database management systems (DBMSs). Confusingly, and not in accordance with the relevant International Organization for Standardization (ISO) standards, the authors limit their use of the term "maintenance" to corrective maintenance for the restoration of the physical structure of a database that was affected adversely during its normal expected use-reading, replacing, inserting, or deleting data-without affecting the associated logical structure. Yet, the focus of this entire survey paper is online database system maintenance, including corrective maintenance. Based on what Sockut and Iyer regard as issues in online database reorganization, the survey is structured into three categories of reorganization: maintenance (their definition), physical restructuring, and logical restructuring. The summary of each reorganization category is presented in a format that can also be read as a tutorial. The authors are careful to note the effects of different kinds of databases-such as relational and hierarchical-as they report their survey of the literature. They are also alert to side effects, such as slower performance; the possible need for storage garbage collection; the additional storage space required; the use of log files; how up-to-date the database is; the effects on indexes, where relevant; and the kind and amount of general additional overhead created while processing online reorganization and providing the capability for processing online reorganization. Two outstanding features of the survey are: first, its comprehensive scope-the authors cover so much of the scattered literature that readers will be saved weeks of research; and, second, the terse and able summaries of the literature, which draw attention to the growing importance of the online system maintenance of online databases. Overall, this paper is a very worthwhile contribution to software maintenance and evolution literature. Online Computing Reviews Service