Shared Memory Failure Indication for Microsecond Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computing systems face challenges in efficiently and quickly handling system failures, leading to potential downtime due to the limitations of existing failover mechanisms, which are often slower than needed to maintain continuous functionality.
Innovation Solution
The implementation of a shared memory system where nodes can store failure indications, allowing for rapid system function takeover by other nodes, leveraging faster memory access compared to traditional alert mechanisms like network broadcasts, enabling microsecond-level failover processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If traditional alert mechanisms like network broadcasts are used for failure detection, then system coverage can be achieved, but failover speed is slow and cannot meet continuous functionality requirements
Solution Approach 1:
The patent introduces shared memory as an intermediary mechanism between failure detection and failover execution. Instead of using slow network broadcasts, the failing node writes a failure indication directly to shared memory, which is immediately accessible to other nodes, thus achieving fast failover while maintaining system reliability
Solution Approach 2:
The patent replaces the mechanical/network-based alert broadcast system with a memory-based indication system. By substituting network communication with direct memory access, the system achieves microsecond-level failover speed while ensuring continuous functionality through reliable failure detection
2Loss of time
If network broadcasts are used for failure alerts, then failure detection can be achieved across the system, but the processing time is too slow for microsecond-level failover
Solution Approach 1:
Shared memory serves as an intermediary that eliminates the need for slow network broadcast transmission. The failure indication is written directly to shared memory, which is immediately visible to other nodes, reducing alert transmission time to memory access speed and enabling microsecond-level failover
Solution Approach 2:
The patent uses the concept of copying the failure indication to a shared location (shared memory) that all nodes can access. This allows any node to immediately detect the failure without waiting for network broadcast propagation, significantly reducing downtime while maintaining fast alert transmission
3Speed
If shared memory is used for failure indication storage, then failover speed increases to microsecond level, but system complexity increases due to shared memory management
Solution Approach 1:
The patent segments the failover system into distinct functional components: failure detection logic, shared memory access interface, and failover execution logic. This segmentation allows each component to be independently optimized and managed, reducing overall system complexity while maintaining microsecond-level failover speed
Solution Approach 2:
The shared memory system is designed with universal access interfaces that can be used by multiple nodes simultaneously. By creating a multi-functional shared memory structure that serves both as failure indication storage and as a coordination mechanism, the patent reduces the need for separate complex communication protocols, thereby reducing device complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In some examples, a node of a computing system may include a failure identification engine and a failure response engine. The failure identification engine may identify a failure condition for a system function of the node and the failure response engine may store a failure indication in a shared memory to trigger takeover of the system function by a different node of the computing system.