Topology Manager for Failure Detection in Distributed Systems

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Distributed computing systems face inefficiencies in reestablishing communication after network or node failures, particularly in large-scale systems where conventional heartbeat-based methods create significant network traffic and load, and fail to quickly determine whether a node is temporarily unavailable or permanently down.

Innovation Solution

A topology manager is designated to redundantly store node information and states, using a timeout-based election protocol to ensure leader presence, and upon notification of a failed node, it retires the node and rebalances data partition replicas among surviving nodes, with load-balancing algorithms considering node properties and resource utilization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional heartbeat-based methods are used for failure detection, then nodes can detect failures, but network traffic and system load increase significantly

Engineering Contradiction:
Improvefailure detection capabilityVSAvoidnetwork traffic
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent extracts the failure detection function from the conventional heartbeat mechanism by introducing a separate topology manager component. This topology manager consolidates failure detection logic and uses efficient timeout-based election protocols instead of continuous heartbeats, reducing network traffic while maintaining reliable failure detection capability.

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If conventional failure detection methods are used, then failures can be detected, but the system cannot quickly determine whether a node is temporarily unavailable or permanently down

Engineering Contradiction:
Improvefailure detectionVSAvoidfailure determination time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by having the topology manager pre-establish timeout thresholds and election protocols before failures occur. When a node becomes unresponsive, the system can quickly determine permanent failure versus temporary unavailability by checking against pre-configured timeout criteria, eliminating the need for prolonged detection delays.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If the system retires failed nodes and rebalances data partitions, then data consistency is maintained, but network load and processing overhead increase

Engineering Contradiction:
Improvedata consistencyVSAvoidnetwork load
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The topology manager serves as an intermediary that coordinates node retirement and data partition rebalancing operations. By centralizing this coordination function, the system can optimize rebalancing operations to minimize network load, using the topology manager's global view to make efficient redistribution decisions rather than having nodes independently negotiate rebalancing.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentUS10341168B2Topology manager for failure detection in a distributed computing system
Publication Date: 2019.07.02 SAP SE
  • US10341168B2 patent drawing
  • US10341168B2 patent drawing

AI summary

A topology manager implemented on a data partition of a distributed computing system can be notified that a destination computing node in the distributed computing system is not responding to a communication request. Upon determining that the destination computing node is dead and/or has a loss of communication with one or more, and optionally a majority of other computing nodes in a plurality of computing nodes of the distributed computing system, the topology manager can retire the destination computing node and cause a load balancing of replicas of data partitions in the distributed computing system to compensate for loss of the retired computing node.