Process Checkpointing for Data System Restart
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data processing systems face significant downtime and resource consumption when terminating and restarting due to system failures, requiring manual reinitialization by specialists and involving separate processes to detect failures, which is inefficient and time-consuming.
Innovation Solution
Implementing a method where individual processes in a multi-process system are informed to suspend processing until a failed connection is restored, allowing execution to resume from the last recorded checkpoint state without shutting down the entire system, using checkpointing techniques to store current states and propagate abort/suspend messages through processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire system is terminated and restarted after a failure, then system reliability is improved through complete reset, but system downtime and resource consumption increase significantly
Solution Approach 1:
The patent segments the monolithic system restart operation into individual process-level operations. When a failure occurs, only the affected process is terminated and restarted, while other processes continue running. This is achieved through process-level state management where each process maintains its own state independently, allowing selective recovery without system-wide shutdown.
Solution Approach 2:
The patent implements preliminary checkpointing where process states are saved at predetermined intervals before failures can occur. When a failure is detected, the system can quickly restore to the most recent checkpoint without needing to perform a complete system restart, significantly reducing recovery time while maintaining reliability.
2Stability of the object's composition
If manual reinitialization by specialists is required after system failure, then system stability is improved through expert intervention, but operational efficiency and productivity decrease
Solution Approach 1:
The patent implements self-service through automated failure detection and recovery mechanisms. The system automatically detects process failures, retrieves saved checkpoint states, and restarts affected processes without requiring manual intervention from specialists. This maintains system stability through structured recovery procedures while dramatically improving operational efficiency by eliminating human response time.
3Difficulty of detecting and measuring
If separate processes are used to detect failures and inform specialists, then failure detection capability is improved, but system complexity and resource consumption increase
Solution Approach 1:
The patent merges the failure detection function into the existing process monitoring infrastructure rather than using separate dedicated detection processes. The same process monitoring mechanisms that track process health also detect failures and trigger recovery procedures, eliminating the need for separate detection processes and reducing overall system complexity while maintaining detection capability.
4Reliability
If the entire system is shut down for restart, then complete system state reset is achieved improving reliability, but processing resource consumption increases during restart
Solution Approach 1:
The patent segments the system restart operation to affect only individual processes rather than the entire system. Resources are consumed only by the specific processes that need restarting rather than all system processes, significantly reducing processing resource consumption during recovery while maintaining reliability through targeted process restoration.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques are disclosed that include a computer-implemented method including transmitting a message (604) in response to a predetermined event (506) through a process stage including at least first and second processes being executed as one or more tasks, the message instructing the abortion of the executing of the one or more tasks, and initiating abortion of execution (606) of the one or more tasks by the one or more of the processes on receiving the messages.