Network Interface Card Anomaly Detection for Faster Path Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing NVMe-oF-based storage systems rely on a heartbeat timeout mechanism for path switching, leading to prolonged delays and reduced reliability and continuity of storage services when an anomaly occurs on a node.
Innovation Solution
A network interface card actively detects anomaly events on nodes and sends notification messages to the host, reducing path switching delays by proactively informing the host to switch to an alternate path.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a heartbeat timeout mechanism is used to detect node anomalies, then the system can detect failures, but the path switching delay becomes excessively long
Solution Approach 1:
The network interface card performs preliminary anomaly detection on nodes before failures occur. By continuously monitoring node status and detecting anomalies proactively, the system can trigger path switching before complete failure occurs, significantly reducing the switching delay compared to waiting for heartbeat timeouts.
2Device complexity
If the host passively waits for heartbeat messages to discover faults, then the system structure remains simple, but service continuity is severely impacted
Solution Approach 1:
The network interface card implements an active feedback mechanism where it continuously monitors node status and immediately notifies the host when anomalies are detected. This feedback loop enables the host to respond promptly to node failures, maintaining storage service continuity without requiring complex passive waiting mechanisms.
3Measurement precision
If heartbeat connections are established between host and node for monitoring, then node status can be tracked, but the response time to anomalies remains unacceptably slow
Solution Approach 1:
The network interface card serves as an intermediary between the node and the host. It actively monitors node status through its direct connection to the node and immediately communicates anomaly information to the host, eliminating the need for the host to continuously check node status through heartbeat connections. This intermediary role significantly accelerates anomaly detection and response speed.
Data Source
AI summary
A node anomaly event processing method is applied to a network interface card in a storage device. The storage device further includes a plurality of nodes configured to manage a storage. The network interface card is communicatively connected to a first node in the plurality of nodes. When detecting an anomaly event related to the first node, the network interface card can actively send a notification message to a host to notify the host that an anomaly occurs on a path on which the first node is located, so that the host performs path switching.