A Fine-Grained Damage Management Scheme in a Self-Healing PostgreSQL System

Abstract

Database Damage Management is an important issuein today's E-Business world. Self-healing capability is an indispensable part of the modern database system to assure high level data availability and integrity for misssion/life/business applications. Although a number of research projects have been done to tackle the emerging data damage threats to DBMS, existing mechanisms are still limited in finely distinguishing uncontaminated data from true corrupted data, which causes availability loss. In this paper, we propose TRACE-FG, a fine-grained zero-systemdown-time database damage tracking, quarantine, and recovery solution with negligible run time overhead. TRACE-FG is able to decompose the "in-danger" unit and save the legitimated work. We built TRACE-FG into the kernel of PostgreSQL. Our experimental results demonstrated that TRACE-FG outperforms the original TRACE system and the standard recovery mechanism in terms of data availability and system service delay.