Multi-Architecture Server Clusters for Scalable Virtual Network Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network architectures face bottlenecks in scalability, deployment costs, and operational efficiency due to the exponential increase in bandwidth demand driven by video and IoT applications, necessitating improved hardware performance and orchestration in network function virtualization (NFV) frameworks.
Innovation Solution
A cloud infrastructure with a first server architecture for high-performance network functions and a second server architecture for non-throughput-dependent functions, coupled via a tunnel connection to jointly service packet flows, utilizing a hierarchical quality of service (HQoS) and resource orchestration to optimize network function processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional hardware-based network solutions are used, then hardware performance is maintained, but scalability and deployment cost efficiency deteriorate
Solution Approach 1:
The patent replaces traditional hardware-based network solutions with software-defined network functions. Network functions are virtualized and implemented as software instances that can be deployed on standard servers, eliminating the need for specialized hardware and enabling scalable, cost-effective deployment while maintaining performance through software optimization.
Solution Approach 2:
The patent implements a universal server architecture that can host multiple different network functions through software. A single server infrastructure can deploy various network functions (firewall, router, load balancer, etc.) as virtual instances, replacing the traditional one-to-one hardware-to-function mapping and enabling flexible, scalable deployment.
2Ease of manufacture
If network functions are virtualized on uniform servers, then deployment cost decreases, but performance for throughput-intensive functions deteriorates
Solution Approach 1:
The patent introduces differentiated server architectures tailored to specific network function requirements. Throughput-intensive functions (like routing and forwarding) are deployed on optimized servers with dedicated hardware accelerators and high-performance networking components, while stateless functions use standard servers. This local optimization ensures each function receives the hardware resources it needs without over-provisioning the entire infrastructure.
Solution Approach 2:
The patent segments the server infrastructure into different architectural types based on function requirements. The system maintains a catalog of server architectures and selects appropriate types for each network function deployment, separating throughput-critical workloads from stateless workloads to optimize both performance and cost efficiency.
3Adaptability or versatility
If multiple server architectures are integrated, then functional versatility improves, but system complexity increases
Solution Approach 1:
The patent introduces a network function orchestrator as an intermediary layer that manages the complexity of multiple server architectures. The orchestrator abstracts the underlying hardware diversity, providing a unified interface for deploying and managing network functions. It automatically selects appropriate server architectures based on function requirements and handles resource allocation, shielding users from architectural complexity while enabling functional versatility.
Solution Approach 2:
The patent uses configurable server architecture templates where hardware and software parameters can be adjusted based on function requirements. Rather than managing completely different architectures, the system modifies parameters (CPU types, memory configurations, hardware accelerator options) within standardized templates, reducing complexity while maintaining versatility.
4Reliability
If hardware resources are allocated for peak demand, then service reliability improves, but resource waste increases
Solution Approach 1:
The patent implements dynamic resource allocation where server resources can be dynamically added or removed based on actual demand. The system monitors network function performance and traffic patterns, automatically scaling compute and storage resources up during peak demand and down during low-demand periods. This dynamic adjustment maintains service reliability when needed while eliminating resource waste during off-peak times.
Solution Approach 2:
The patent pre-provisions server resources based on predicted demand patterns and service level agreements. Rather than reacting to peak demand, the system uses historical data and traffic forecasts to allocate resources in advance, ensuring availability during expected peak periods while avoiding over-provisioning during low-demand periods. This predictive approach balances reliability and resource efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a method and a network apparatus in a cloud infrastructure. The network apparatus includes a processor, and a memory coupled to the processor, the memory for storing computer instructions that, when executed by the processor, cause the processor to generate a resource abstraction layer relating a network cluster including a first server architecture with a first resource set and a second server architecture with a second resource set, to generate an orchestration layer configured to receive a packet flow request and, in response to the packet flow request, schedule virtual network resources based on the resource abstraction layer for servicing a packet flow, and to deploy the virtual network resources to receive and service the packet flow.