Service Availability

Database management systems are critical components of highly available appli- tions. To meet this need, many highly available database management systems have been developed. This paper describes the architectures that are internally used to construct these highly available databases. These architectures are examined from the perspective of both process redundancy and logical data redundancy. Process red- dancy is always required; it refers to the maintenance of redundant database processes that can take over in case of failure. Data redundancy is also required. Data red- dancy can be provided at either the physical or the logical level. Although both forms of data redundancy can provide high availability, this paper has concentrated on lo- cal data redundancy since that is a case where the database explicitly manages the data copies. We believe that process and data redundancy are useful means to describe the availability characteristics of these software systems. References 1. Application Interface Specification, SAI-AIS-A. 01. 01, April 2003. Service Availability Forum, available at www. saforum. org. 2. Gray, J. and Reuter, A. : Transaction Processing Systems, Concepts and Techniques. M- gan Kaufmann Publishers, 1992. 3. How MySQL Cluster Supports 99. 999% Availability. MySQL Cluster white paper, MySQL AB, 2004, available at mysql. com/cluster/. 4. Hu, K. , Mehrotra, S. , Kaplan, S. M. : Failure Handling in an Optimized Two-Safe Approach to Maintaining Primary-Backup Systems. Symposium on Reliable Distributed Systems 1998: 161-167. 5. Humborstad, R. , Sabaratnam, M. , Hvasshovd, S-O. , Torbjørnsen, Ø.

Contenu
ISAS 2004.- Architecture of Highly Available Databases.- Data Persistence in Telecom Switches.- Distributed Redundancy or Cluster Solution? An Experimental Evaluation of Two Approaches for Dependable Mobile Internet Services.- OpenHPI: An Open Source Reference Implementation of the SA Forum Hardware Platform Interface.- Quality of Service Control by Middleware.- Benefit Evaluation of High-Availability Middleware.- A Measurement Study of the Interplay Between Application Level Restart and Transport Protocol.- Service-Level Management of Adaptive Distributed Network Applications.- A Methodology on MPLS VPN Service Management with Resilience Constraints.- Highly Available Location-Based Services in Mobile Environments.- On Enhancing the Robustness of Commercial Operating Systems.- A Modular Approach for Model-Based Dependability Evaluation of a Class of Systems.- Rolling Upgrades for Continuous Services.- First Experience of Conformance Testing an Application Interface Specification Implementation.- On the Use of the SA Forum Checkpoint and AMF Services.