Kubernetes Edge Node Datapath Split for L4 and L7 Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current edge node architectures based on monolithic appliance models, such as those using Data Plane Development Kit (DPDK), face limitations when handling networking and security functions beyond Layer 4 (L4) services, particularly in scaling and resource management for Layer 5-7 (L5-L7) services in Kubernetes environments.
Innovation Solution
Implementing a Kubernetes cluster with multiple nodes that execute pods, where a first pod handles L4 operations and separate pods handle L5-L7 services for logical routers, utilizing autoscaling to dynamically adjust node resources and distribute workload across nodes, enabling efficient processing of data messages between logical and external networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a monolithic appliance model using DPDK is used for edge nodes, then high performance networking is achieved for L4 services, but scalability and resource management are limited for L5-L7 services
Solution Approach 1:
The patent segments the monolithic edge appliance into separate Kubernetes pods: a DPDK pod for high-performance L4 networking functions and additional pods for L5-L7 services. This segmentation allows each service layer to be independently managed, scaled, and optimized, resolving the contradiction by maintaining high performance for L4 while enabling scalability for L5-L7 services through modular architecture
Solution Approach 2:
The Kubernetes platform provides universal resource management and orchestration capabilities that serve multiple service types (L4, L5, L6, L7) through a common infrastructure. The shared control plane and resource pool enable the system to handle diverse networking functions with unified management, achieving both high performance and broad adaptability across different service layers
2Adaptability or versatility
If networking functions are moved to application layer (L5-L7), then service functionality is enhanced, but the monolithic architecture shows limitations in handling these functions
Solution Approach 1:
Kubernetes serves as an intermediary layer between the DPDK data plane and the L5-L7 application services. It provides standardized interfaces for service deployment, resource allocation, and orchestration, enabling complex networking functions to be managed through a simplified control plane while maintaining the high-performance DPDK backend, thus enhancing service functionality without proportionally increasing operational complexity
3Adaptability or versatility
If separate pods are used for each logical router, then resource utilization and scalability are improved, but workload distribution and node management become more complex
Solution Approach 1:
Kubernetes implements self-service mechanisms through its built-in scheduling, load balancing, and resource management systems. The platform automatically distributes pods across nodes based on resource availability, handles service discovery and networking, and manages lifecycle operations, thereby improving resource utilization while reducing the operational burden of workload distribution through automation
Data Source
AI summary
Some embodiments of the invention provide a system for implementing multiple logical routers. The system includes a Kubernetes cluster that includes multiple nodes, with each node executing a set of pods. The set of pods include a first pod for performing a first set of data message processing operations for the multiple logical routers and at least one respective separate pod for each respective logical router of the multiple logical routers. Each respective pod is for performing a respective second set of data message processing operations for the respective logical router.