CloudRAN Failure Detection via Job Flow Measurement Reports
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional heartbeat/watchdog mechanisms for high availability (HA) detection in CloudRAN are ineffective due to their inability to distinguish between node failures and network failures, and they increase system load with frequent heartbeat packets, which is not suitable for the millisecond-level data packet intervals in CloudRAN.
Innovation Solution
A method is introduced where processing nodes in a cloud environment send measurement reports to a management node, providing information about jobs pulled from and pushed to upper and under layers, allowing for failure detection based on these reports, and a voting mechanism is used to differentiate between node and network failures, reducing the need for frequent heartbeat packets and avoiding processing stalls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional heartbeat/watchdog mechanism is used for HA detection, then failure detection capability is provided, but system load increases due to frequent heartbeat packets
Solution Approach 1:
The patent extracts the failure detection function from the conventional heartbeat mechanism by introducing measurement reports that contain job status information. Instead of relying solely on periodic heartbeat packets, the system extracts meaningful failure indicators from actual job processing data, thereby reducing the need for frequent dedicated heartbeat transmissions and lowering system load while maintaining detection capability.
Solution Approach 2:
The measurement report mechanism serves multiple functions: it provides failure detection, monitors job status, and tracks data flow between layers. By making the measurement report multi-functional, the patent eliminates the need for separate heartbeat packets, thereby reducing system load while maintaining or improving failure detection capability.
2Reliability
If conventional heartbeat mechanism is used, then failure detection is enabled, but inability to distinguish between node failures and network failures occurs
Solution Approach 1:
The patent implements feedback through measurement reports that contain detailed job status information from multiple layers. By analyzing whether upper layer jobs are being pulled and whether under layer jobs are being pushed, the system receives feedback that enables precise determination of failure type - distinguishing between node failures (no job movement) and network failures (job movement patterns indicate connectivity issues).
Solution Approach 2:
The patent adds another dimension to failure detection by incorporating job status information from multiple layers (upper and under layers) into the measurement reports. This multi-dimensional data allows the system to differentiate failure types by analyzing job flow patterns across different layers, rather than relying on single-dimension heartbeat signals.
3Reliability
If frequent heartbeat packets are sent for HA detection, then failure detection speed is improved, but processing stalls occur in CloudRAN environment
Solution Approach 1:
The patent ensures continuity of useful action by making measurement reports an integral part of the normal job processing flow rather than separate heartbeat interruptions. Nodes continuously generate measurement reports as part of their job execution, eliminating the need for dedicated heartbeat packets that could cause processing stalls while maintaining continuous monitoring capability.
Data Source
AI summary
Method, management node and processing node are disclosed for continuous availability in a cloud environment. According to an embodiment, the cloud environment comprises a plurality of layers and each layer includes at least two processing nodes. Each processing node in a layer can pull job(s) from the processing nodes in the upper layer if any and prepare job(s) for the processing nodes in the under layer if any. A method implemented at a management node comprises receiving measurement reports from the plurality of layers. The measurement report of each processing node comprises information about job(s) pulled from the upper layer if any and job(s) pulled by the under layer if any. The method further comprises determining information about failure in the cloud environment based on the measurement reports.