Edge Computing Mesh Registry Agents for Bandwidth Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Edge computing systems face bandwidth constraints in mesh networks, leading to inefficient image distribution and high traffic due to the need for images to be downloaded through multiple links, resulting in significant bandwidth demand and potential bottlenecks, especially when using containerization.
Innovation Solution
Distributing registry functions among hosts in a mesh computing system, allowing edge nodes to download images from neighboring hosts instead of a central registry, thereby reducing the need for central registry requests and optimizing image distribution by using registry agents or local registries to manage and supply images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If images are downloaded from a central registry through multiple mesh network links, then image distribution can be achieved, but bandwidth demand increases and bottlenecks occur
Solution Approach 1:
The patent segments the centralized registry function into distributed registry agents deployed across multiple hosts in the mesh network. Each registry agent stores and serves images locally, eliminating the need for all nodes to download images through multiple links from a central registry. This segmentation reduces bandwidth demand while maintaining image distribution capability.
Solution Approach 2:
The patent introduces a new dimensional approach by adding registry agents at the edge nodes and intermediate hosts within the mesh network. Instead of a single-point central registry architecture, the system creates a multi-dimensional distribution network where images can be obtained from multiple nearby sources, reducing the path length and bandwidth consumption for image downloads.
2Device complexity
If a central registry is used for image distribution, then image management is simplified, but traffic in the mesh network increases
Solution Approach 1:
The patent introduces registry agents as intermediary components between the central registry and edge nodes. These agents cache images locally and serve requests from neighboring nodes, acting as intermediaries that reduce traffic to the central registry while maintaining centralized coordination for image updates and management.
Solution Approach 2:
The patent implements preliminary action by having registry agents pre-cache images locally before they are needed by other nodes. When a node requires an image, the registry agent can serve it immediately from local storage rather than requiring a download through the mesh network, thereby reducing network traffic while maintaining simplified central registry management.
3Extent of automation
If images are obtained from distant central registry, then centralized control is maintained, but bandwidth bottlenecks occur on backhaul links
Solution Approach 1:
The patent applies local quality by enabling each registry agent to serve images from its local storage capacity, allowing nearby nodes to obtain images from nearby sources rather than all nodes relying on distant central registry access. This maintains centralized control through coordination protocols while locally reducing bandwidth consumption on backhaul links.
Data Source
AI summary
A method of traffic reduction in a mesh computing system (400), the mesh computing system (400) comprising hosts located on edge nodes of the mesh computing system (400) and a central registry located outside the mesh computing system (400), the central registry holding the images. The method comprises, at a first host located at a first edge node, receiving (920) a request from a client for an image, sending (930) a request for the image to at least one other host of the mesh computing system (400). When the first host receives (940) notification that at least a second host holds the image, the first host downloads (960) the image from the second host to the first host. The first host creates (970) a container from the image. A host at a node (636; 700) and a mesh computing system (400) are also provided.