Prefetch Engine for Distributed Graph Hub Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large graphs, especially distributed graphs, experience significant delays due to high communication costs and resource inefficiencies caused by node sampling and data transmission between devices.
Innovation Solution
The method involves logically dividing a graph into subgraphs stored on interconnected devices, with hub nodes connecting adjacent subgraphs, and using a prefetch engine to proactively load attributes and node identifiers from one device to another, especially when the gap between hub nodes is a single hop, thereby reducing latency and communication overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If graph data is distributed across multiple devices for processing, then the system can handle large-scale graphs with billions of nodes, but communication latency and bandwidth consumption increase significantly
Solution Approach 1:
The patent implements a prefetching mechanism that proactively loads node attributes and hub node information from remote devices before they are actually needed for processing. When a node is accessed, the system automatically prefetches its attributes and information about connected hub nodes, so that this data is already available in local memory when needed, eliminating communication latency during the actual processing phase
Solution Approach 2:
The patent introduces hub nodes as intermediary elements that connect different subgraphs distributed across devices. Each device stores information about hub nodes on other devices, creating an indexed map that enables efficient routing and prefetching. This intermediary structure reduces the complexity of direct peer-to-peer communication across the distributed system
2Productivity
If node sampling is performed in distributed graphs, then relevant data can be extracted efficiently, but the communication cost and overall delay increase due to data transmission between devices
Solution Approach 1:
The patent optimizes the distributed graph structure by identifying and marking hub nodes in each subgraph. Each device maintains local information about which nodes are hub nodes and stores their attributes locally. This local quality optimization means that commonly accessed hub node information is cached locally, reducing the need for repeated cross-device communications during sampling operations
3Stability of the object's composition
If hub node information is stored on remote devices, then data distribution is optimized, but access latency increases when nodes are sampled
Solution Approach 1:
The system performs preliminary loading of hub node attributes and connected node information into local memory before they are needed. When processing begins, the prefetch engine has already populated local caches with frequently accessed hub node data, so that subsequent access operations can proceed at high speed without remote communication delays
Solution Approach 2:
The patent creates local copies of hub node information and attribute data on each device. Instead of maintaining a single source of truth that requires remote access, each device maintains cached copies of the hub node information it needs, allowing fast local access while the original data remains distributed across the system
Data Source
AI summary
Various embodiments of the present disclosure relate to a computer-implemented method, a system, and a storage medium, where a graph stored in a computing system is logically divided into subgraphs, the subgraphs are stored on different interconnected (or coupled) devices in the computing system, and nodes of the subgraphs include hub nodes connected to adjacent subgraphs. Each device stores attributes and node structure information of the hub nodes of the subgraphs into other devices, and software or hardware prefetch engine on the device prefetches attributes and node structure information associated with a sampled node. A prefetcher on a device interfacing with the interconnected (or coupled) devices may further prefetch attributes and node structure information of nodes of the subgraphs on other devices. A traffic monitor is provided on an interface device to monitor traffic. When the traffic is small, the interface device prefetches node attributes and node structure information.