This paper provides a terminological framework for describing transaction-oriented recovery schemes in database systems, focusing on conceptual rather than implementation-dependent aspects. The authors introduce terms such as "materialized database," "propagation strategy," and "checkpoint" to classify various implementations uniformly. They also present a classification scheme for logging techniques, which are defined using these terms. The paper aims to establish precise terminology for a topic where confusion and implementational details often complicate understanding. The introduction covers the expected functions of database recovery, the types of failures to be anticipated, and the concept of consistency. The paper then delves into the mapping hierarchy of a DBMS, discussing the objects and operations at different levels of abstraction. It also explores the storage hierarchy and different views of a database, including the current, materialized, and physical databases. The concept of propagation is introduced, distinguishing between direct and indirect page allocation schemes, and their impact on recovery. The paper concludes with a detailed discussion of crash recovery, including the state of the database after a crash, the types of log information required, and examples of recovery techniques. The authors evaluate these techniques based on criteria such as speed, scope of recovery, and amount of redundant information needed.This paper provides a terminological framework for describing transaction-oriented recovery schemes in database systems, focusing on conceptual rather than implementation-dependent aspects. The authors introduce terms such as "materialized database," "propagation strategy," and "checkpoint" to classify various implementations uniformly. They also present a classification scheme for logging techniques, which are defined using these terms. The paper aims to establish precise terminology for a topic where confusion and implementational details often complicate understanding. The introduction covers the expected functions of database recovery, the types of failures to be anticipated, and the concept of consistency. The paper then delves into the mapping hierarchy of a DBMS, discussing the objects and operations at different levels of abstraction. It also explores the storage hierarchy and different views of a database, including the current, materialized, and physical databases. The concept of propagation is introduced, distinguishing between direct and indirect page allocation schemes, and their impact on recovery. The paper concludes with a detailed discussion of crash recovery, including the state of the database after a crash, the types of log information required, and examples of recovery techniques. The authors evaluate these techniques based on criteria such as speed, scope of recovery, and amount of redundant information needed.