Reliability Mechanisms for ADAMS

Authors:Son, Sang, Institute for Parallel ComputationUniversity of Virginia Pfaltz, John, Institute for Parallel ComputationUniversity of Virginia

The goal of checkpointing in database management systems is to save database states on a separate secure device so that the database can be recovered when errors and failures occur. This paper presents a non-interfering checkpointing mechanism being developed for ADAMS. Instead of waiting for a consistent state to occur, our checkpointing approach constructs a state that would result by completing the transactions that are in progress when the global checkpoint begins. The checkpointing algorithm is executed concurrently with transaction activity while constructing a transaction-consistent checkpoint on disk, without requiring the database quiesce. This property of noninterference is highly desirable to real-time applications, where restricting transaction activity during the checkpointing operation is in many cases not feasible. Two main properties of this checkpointing algorithm are global consistency and reduced interference, both of which are crucial for achieving high availability.
Note: Abstract extracted from PDF text

All rights reserved (no additional license for public reuse)
Source Citation:

Son, Sang, and John Pfaltz. "Reliability Mechanisms for ADAMS." University of Virginia Institute for Parallel Computation Tech Report (1988).

University of Virginia, Institute for Parallel Computation
Published Date: