Node State Snapshot Restoration for Parallel Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In parallel computer systems, jobs often leave nodes in an unstable state with orphaned processes and resources un reclaimed, leading to inefficient job execution and potential failures, as rebooting is time-consuming and may not restore the node to a known state reliably.
Innovation Solution
Storing a snapshot of the node's system state allows for efficient restoration to a known state, including determining the storage location and compression of the snapshot, enabling faster and more reliable node state recovery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a node is rebooted to restore it to a known state, then the node can be returned to a stable state, but the process is time-consuming and may not reliably restore the node to a known state
Solution Approach 1:
The patent creates a snapshot (copy) of the node's system state at a known good point. This snapshot includes memory contents, register states, and other system information. Instead of rebooting the entire node, the system restores the saved snapshot data back into memory, effectively copying the known good state back into the running system. This approach is faster and more reliable than rebooting because it directly restores the exact known state without the variability introduced by the asynchronous boot process.
2Productivity
If a snapshot of the system state is stored and used for restoration, then node state restoration is faster and more reliable, but additional storage and compression management is required
Solution Approach 1:
The system performs preliminary actions by capturing and storing a snapshot of the system state at a known good point before jobs are executed. This snapshot is saved to storage with associated metadata including compression information. When restoration is needed, the pre-prepared snapshot can be quickly loaded and applied, eliminating the need for time-consuming state capture during restoration operations.
Solution Approach 2:
The patent changes the state parameters of the snapshot by compressing the saved system state data. The compression reduces the storage requirements and speeds up transfer operations. The system manages compression parameters and can adjust them based on available storage resources and performance requirements, optimizing the balance between storage efficiency and restoration speed.
Data Source
AI summary
In one aspect of the invention, a method is provided. The method may include: (1) storing a snapshot of a system state of a node; (2) executing a job on the node; and (3) restoring the node to the system state using the stored snapshot of the system state.