Pre-crash Resource Dumping for Faster Crash Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dumping system state information and memory in computing devices are inefficient, especially in large systems, as they involve recording both static and non-static resources after a crash, leading to prolonged recovery times due to the inclusion of unnecessary static data.
Innovation Solution
Implementing a system with identification, determination, and dumping engines that identify and dump only changed resources prior to a crash, using idle resources to filter and determine changes without impacting performance, thereby reducing the time required for dumping and recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all system state information and memory are recorded after a crash, then complete crash analysis is achieved, but recording time and recovery time are excessively prolonged
Solution Approach 1:
The system performs preliminary identification and dumping of resources before a crash occurs. The identification engine continuously monitors and identifies resources, and the dumping engine dumps these resources to storage before the crash, so that when a crash occurs, the analysis can be performed quickly on already-prepared data without the need to dump resources after the crash.
Solution Approach 2:
The system extracts only the necessary dynamic resources that are relevant to crash analysis, separating them from static resources that do not contribute to crash diagnosis. The determination engine identifies which resources have changed and are relevant to the crash, and only these resources are dumped to storage, eliminating the need to process all system state information and memory.
2Loss of information
If all resources including static data are dumped, then comprehensive system state is captured, but the amount of data to be processed increases significantly
Solution Approach 1:
The determination engine extracts only the dynamic resources that have changed since the last dump operation, separating them from static resources. This extraction process identifies resources that are actually relevant to crash analysis, such as modified memory regions, changed system state information, and updated configuration data, while excluding static resources that remain unchanged and provide no diagnostic value.
Solution Approach 2:
The system applies different treatment to different types of resources based on their characteristics. Dynamic resources that have changed are identified and dumped with high priority, while static resources are excluded from the dump operation. This local quality approach ensures that each resource is handled according to its specific nature and relevance to crash analysis.
3Loss of time
If resource dumping is performed continuously to capture changes, then real-time crash data is available, but system performance is impacted
Solution Approach 1:
The system performs resource identification and dumping operations periodically or at scheduled intervals rather than continuously. The identification engine monitors resources and triggers dump operations based on time intervals or specific conditions, allowing the system to balance between having up-to-date crash data and maintaining normal system performance by not constantly monitoring and dumping resources.
Data Source
AI summary
Dumping resources can include identifying a resource associated with a computing device, dumping the resource associated with the computing device, determining whether the resource has changed since the resource has been dumped, and dumping the resource, prior to a crash of the computing device occurring, when the resource has changed since the last dump.