In this paper, a terminological framework is provided for describing different transaction-oriented recovery schemes for database systems in a conceptual rather than an implementation-dependent way. By introducing the terms materialized database, propagation strategy, and checkpoint, we obtain a means for classifying arbitrary

Abstract. Techniques similar to shadow pages have been suggested for use in rollback and crash recovery for nested transactions. However, undo/redo log meth-ods have not been presented, though undo/redo logs are widely used for transaction recovery, and perhaps preferable to shadow methods. We develop a scheme of log-based recovery for nested transactions. The result-ing design is promising because it requires a relatively small number of extensions to a similar scheme of recov-ery for single-level transactions. 1. Overview Our goal is to develop algorithms for log-based rollback and recovery of nested transactions. We assume general familiarity with transaction oriented concurrency con-trol and recovery; [Gray 78) and [Haerder and Reuter 831 are good introductions to the subject. We also as-sume some familiarity with nested transactions [Moss 81, Moss 82, Moss 85, Moss 861. In presenting the new design, we first review transac-tion commit semantics, for both single-level and nested transactions. We next describe a scheme of recovery for single-level transactions, which is then extended to nested transactions. Finally, we offer our conclusions concerning the results. 2. Transaction Semantics Single-level transactions consist of some number of ac-tions which are to be performed against a database in an atomic fashion. In particular, transactions should have the following characteristics: l Transactions should be integral (atomic): eventually, either all of a transaction’s actions are done, or none of them. This should be true regardless of failures, up to the overall resiliency required of the system. If a trans-action’s actions are performed, it is called committed; if Permission to copy without fee all or part Of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the VLDB copyright notice and the title of the publication and its date appear, and notice is given that copying is

Checkpointing a database is a vital technique to reduce the recovery time in the presence of a failure. For distributed databases, checkpointing also provides an efficient way to perform global reconstruction. However, the need for global reconstruction is infrequent in most distributed databases. In this paper, we survey and classify previous approaches for checkpointing a distributed database. We argue for a less restrictive and less resource-consuming approach to checkpoint distributed databases in an integrated distributed database system. We also argue against the use of any type of global checkpointing for a federated or multidatabase system.