Cluster Request Timeout Detection for Long-Latency Task Switching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In distributed cluster systems, as the number of computing nodes increases, networking scales become larger, leading to long-latency requests that cause computing nodes to wait for responses, resulting in resource waste and reduced utilization.

Innovation Solution

Implement a detection unit in each computing node to monitor request response times and send a message when a request times out, allowing the processing unit to switch to another task, thereby avoiding prolonged waiting and resource waste.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the quantity of computing nodes in the distributed cluster system increases, then computing performance and system scale are improved, but networking scale becomes larger and request response time increases

Engineering Contradiction:
Improvecomputing performanceVSAvoidrequest response time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by sending a notification message to the computing node before the request actually times out. The system predicts potential timeout based on historical response time data and network conditions, and proactively notifies the computing node to switch tasks before the actual timeout occurs, thus preventing resource waste while maintaining system scalability.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If the computing node waits for a response to the request before performing computing or another action, then request processing completeness is ensured, but computing resources are wasted and utilization decreases

Engineering Contradiction:
Improverequest processing completenessVSAvoidcomputing node utilization
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent implements feedback by continuously monitoring request response times and comparing them against historical data and thresholds. When a request shows signs of timing out (response time exceeds threshold), the system provides feedback to the computing node through a notification message, enabling the node to adjust its behavior (switch tasks) while maintaining overall system reliability through subsequent request cancellation or retry mechanisms.

Inventive Principle:
Principle #23Feedback

3Device complexity

If the computing node cannot sense whether the request times out, then system simplicity is maintained, but computing resources are seriously wasted

Engineering Contradiction:
Improvesystem complexityVSAvoidcomputing resource waste
Core Design Contradiction:
Device complexityVSLoss of energy

Solution Approach 1:

The patent introduces an intermediary component (the notification message mechanism) that bridges the computing node and the request management system. This intermediary proactively communicates timeout predictions to computing nodes without requiring complex timeout detection logic within each node, thus adding minimal complexity while effectively preventing resource waste through task switching.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS20250378139A1Distributed Cluster System and Related Long-Latency Request Processing Method
Publication Date: 2025.12.11 HUAWEI TECH CO LTD
  • US20250378139A1 patent drawing
  • US20250378139A1 patent drawing
  • US20250378139A1 patent drawing

AI summary

A processing unit of a first computing node is configured to send a first request to a second computing node, and a detection unit of the first computing node is configured to, when the first request times out, send a first message to the processing unit of the first computing node. The first message includes one or more of long-latency timeout information and blocked path information. When a response time of the second computing node to the first request is greater than a first threshold, the first request times out. The first threshold is determined based on a plurality of response times, and the plurality of response times are respectively response times of the second computing node to a plurality of requests that have been sent by the first computing node.