Predictive Failure Avoidance
Abstract
The goals of the PFA project are to explore automated techniques for failure avoidance at runtime. The novel approach is to use sophisticated program analyses to: (1) predict when and where failures may happen,(2) synthesize failure avoidance logic that modifies the state of the program to avoid potential failures, and (3) inject monitoring to detect when failures happen at runtime and when they do execute the failure avoidance logic.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2022
- Accession Number
- AD1210804
Entities
People
- Matthew B. Dwyer
- ThanhVu Nguyen
Organizations
- University of Virginia