Container Cluster Communication via Service Broker

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In container management systems like Kubernetes, communicating between clusters requires assigning a public IP address, which raises security concerns and increases costs, especially when the service is not intended to be publicly accessible.

Innovation Solution

The system generates a communication link between clusters without exposing IP addresses by using a domain name service (DNS) and an HTTP(S) master proxy, allowing clusters to communicate securely and reducing IP assignment costs through a load balancer positioned between clusters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a public IP address is assigned to enable communication between clusters, then cluster communication capability is improved, but security is worsened and costs increase

Engineering Contradiction:
Improvecluster communication capabilityVSAvoidsecurity risk
Core Design Contradiction:
Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The patent introduces a service broker as an intermediary component that mediates communication between clusters. The service broker receives service requests from applications, resolves service names to addresses, and forwards requests through appropriate proxies. This intermediary eliminates the need for direct public IP exposure while enabling cluster communication, thus resolving the contradiction between communication capability and security.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If a public IP address is assigned to enable communication between clusters, then cluster communication capability is improved, but costs are increased

Engineering Contradiction:
Improvecluster communication capabilityVSAvoidIP address allocation cost
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent implements a virtual service name system that creates a logical copy/abstraction of service identification. Instead of using physical public IP addresses for service identification, the system uses virtual service names that are resolved through the service broker. This virtualization approach eliminates the need for costly public IP assignments while maintaining communication functionality.

Inventive Principle:
Principle #26Copying

3Ease of operation

If service discovery is implemented across remote clusters, then service accessibility is improved, but network complexity is worsened

Engineering Contradiction:
Improveservice accessibilityVSAvoidnetwork configuration complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The service broker implements self-service functionality by automatically resolving service names to addresses and managing the service discovery process. Applications simply declare their service dependencies using service names, and the service broker autonomously handles the resolution and routing. This automation eliminates the need for manual network configuration and reduces complexity while improving service accessibility.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS10778798B2Remote service access in a container management system
Publication Date: 2020.09.15 HEWLETT PACKARD ENTERPRISE DEV LP
  • US10778798B2 patent drawing
  • US10778798B2 patent drawing
  • US10778798B2 patent drawing

AI summary

A container management system having a first cluster that includes a first service for generating an operation and a second cluster that includes a second service for generating an operation, the second cluster located remotely from the first cluster. A master proxy adjusts a request from the first cluster to access the second service from the second cluster, routes a transmission of the adjusted request to the second cluster and adjusts a response to the adjusted request from the second cluster to the first cluster. The first cluster uses the response to generate a cluster operation based on the first service and the second service.