Provenance Tracking via Binary Instrumentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing provenance tracking systems are limited in their ability to track detailed data lineage across distributed systems, especially in environments where access to internal systems is restricted, such as consumer devices and applications, and lack the capability to automatically embed tracking mechanisms into applications without source code access.
Innovation Solution
A method for tracking provenance information involves creating a provenance layer that instrumented applications can use to log data events, with overwritten library call instructions linking the application to instrumented libraries, enabling the tracking of data operations and generating a provenance log, even in scenarios where the application's source code is not available.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If provenance tracking is implemented by adding a parallel data management system, then provenance tracking capability is achieved, but system complexity and invasiveness increase
Solution Approach 1:
The patent merges provenance tracking functionality directly into the existing application binary by embedding tracking code within the application itself, rather than maintaining a separate parallel system. This integration approach reduces overall system complexity while preserving provenance tracking capabilities.
Solution Approach 2:
The patent introduces an intermediary provenance layer that sits between the application and the underlying storage/system. This layer intercepts and logs data events without requiring modifications to the application's core logic, thereby achieving provenance tracking with minimal invasiveness.
2Ease of operation
If provenance tracking is implemented with less-invasive integration by observing network traffic, then invasiveness is reduced, but tracking scope and detail are limited
Solution Approach 1:
The patent segments the application into instrumented components where specific library calls and data operations are identified and instrumented individually. This allows selective tracking of critical data events without instrumenting the entire application, balancing invasiveness with tracking completeness.
Solution Approach 2:
The patent creates a universal provenance layer that can track multiple types of data events (read, write, delete, transform) across different data types and storage locations through a unified instrumentation mechanism, enabling comprehensive tracking without requiring separate solutions for each scenario.
3Measurement precision
If provenance tracking requires access to application internals and source code, then detailed provenance information can be collected, but adaptability to consumer applications is lost
Solution Approach 1:
The patent performs preliminary instrumentation of the application binary before deployment, embedding the provenance tracking code during the build or deployment phase. This allows the application to be instrumented once, after which it can be deployed to consumer devices without requiring source code access or further modifications at the consumer end.
Solution Approach 2:
The patent creates instrumented copies of the application binary that contain embedded provenance tracking code. These instrumented copies can be distributed and executed on consumer devices without requiring access to the original source code, enabling provenance tracking in consumer applications while maintaining measurement precision.
Data Source
AI summary
A method for tracking provenance information includes catching and logging data events performed by an instrumented application at a provenance layer. Overwritten library call instructions provide library calls to one or more instrumented libraries that invoke the provenance layer to track data operations. A provenance log is created that includes the logged data events.