Deadlock Recovery for Distributed Network Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In cloud computing systems with multiple VLANs and spanning trees, the collective lossless traffic can form loops, leading to deadlock states where network devices become congested and unable to transmit packets, causing a deadlock that existing priority flow control systems cannot resolve.
Innovation Solution
Implementing a deadlock recovery system where network devices enter a deadlock recovery state by dropping queued lossless packets older than a threshold age, reducing memory utilization and allowing resumed packet transmission, thereby alleviating the deadlock.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If priority flow control is implemented in multiple VLANs with lossless traffic, then traffic priority and lossless delivery are improved, but deadlock states occur when collective lossless traffic forms loops
Solution Approach 1:
The patent converts the harmful deadlock state into a beneficial recovery mechanism by introducing a timer that triggers when flow control messages are not received within a expected time window. This timer-based approach transforms the harmful deadlock condition into a controlled recovery process where packets are selectively dropped to break the deadlock and restore normal operation.
Solution Approach 2:
The patent implements feedback mechanisms through flow control messages (pause and resume messages) exchanged between network devices. When a device does not receive expected feedback within a time window, it initiates deadlock recovery by dropping packets. This feedback loop enables the system to detect and respond to deadlock conditions dynamically.
2Reliability
If lossless traffic is buffered at network devices, then packet delivery reliability is improved, but memory utilization increases and deadlock recovery becomes difficult
Solution Approach 1:
The patent applies discarding and recovering by selectively dropping old packets during deadlock recovery while preserving newer packets. The timer-based mechanism allows the system to discard packets that have been queued too long (indicating potential deadlock) while maintaining the ability to recover and resume normal packet delivery once the deadlock is resolved.
3Stability of the object's composition
If multiple spanning trees are used for multiple VLANs, then loop avoidance for individual VLANs is improved, but collective loops can still form causing deadlock
Solution Approach 1:
The patent applies universality by implementing a deadlock recovery mechanism that works across multiple VLANs and spanning trees. The timer-based deadlock detection and recovery process is not VLAN-specific but applies universally to any lossless traffic flow, enabling the system to handle deadlock conditions in complex multi-VLAN environments regardless of the underlying spanning tree configuration.
Data Source
AI summary
A system for deadlock recovery of distributed devices may include a processor and memory. The processor may transmit packets to a device, receive a pause message indicating that the packet transmission should be paused, and initiate a timer and pause the packet transmission in response to receiving the pause message. The processor may enter a deadlock recovery state if the timer reaches a timeout before a resume message is received that indicates that the packet transmission can resume. The processor may, while in the deadlock recovery state, drop packets that have a packet age that is greater than a threshold, and may exit the deadlock recovery state upon dropping a packet that has a packet age less than the threshold, or upon receiving the resume message. The processor may re-initiate the timer if the resume message has not been received, otherwise the processor may resume the packet transmission.