A Survey on Graph Neural Networks for Microservice-Based Cloud Applications

Area of Science:

• Computer Science
• Artificial Intelligence
• Cloud Computing

Background:

• Graph neural networks (GNNs) demonstrate significant success across diverse domains like traffic analysis and computer vision.
• The rise of cloud-native applications has spurred interest in applying GNNs to microservice architecture challenges.
• Existing research explores GNNs for microservice design, prototyping, and large-scale deployment.

Purpose of the Study:

• To provide a comprehensive review of GNN applications in microservice-based systems.
• To identify key areas where GNNs are utilized within microservice architectures.
• To outline future research directions for GNN-based solutions in cloud systems.

Main Methods:

• Systematic literature review of recent studies on GNNs for microservice applications.
• Categorization of GNN applications based on microservice lifecycle stages (design to deployment).
• Analysis of GNN design patterns addressing specific microservice challenges.

Main Results:

• Convolutional graph neural networks (ConGNNs) are currently popular for microservice applications in cloud systems.
• Spatio-temporal graph neural networks (STGNNs) and dynamic graph neural networks (DGNNs) represent emerging trends for advanced studies.
• GNNs offer versatile solutions for various challenges in microservice architecture.

Conclusions:

• GNNs are a powerful tool for optimizing microservice-based cloud-native applications.
• ConGNNs are widely adopted, while STGNNs and DGNNs show promise for future advancements.
• Further research into GNNs can significantly enhance the efficiency and scalability of cloud systems.