Hierarchical Domain Partitioning for Scalable Network Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Scalability issues in computer networks become unmanageable as the number of servers increases, leading to high hardware costs and operational complexity due to large routing tables and NP-complete problems in network traffic management.
Innovation Solution
Partitioning computer networks into distinct physical or overlay domains interconnected in a hardware or software-defined hierarchy, allowing control and forwarding plane functions to be managed independently within each domain, reducing hardware costs and operational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the number of servers increases to improve network capacity and coverage, then network scalability is improved, but hardware costs and operational complexity increase to unmanageable levels
Solution Approach 1:
The patent applies segmentation by dividing the computer network into multiple hierarchical domains (e.g., leaf domains and spine domains). Each domain manages its own routing table independently, so the routing table size in each domain remains manageable even as the overall network scales to millions of servers. This hierarchical segmentation allows the network to scale without proportionally increasing operational complexity at each node.
2Adaptability or versatility
If the number of servers increases to improve network capacity, then network scalability is improved, but routing table size exceeds memory capacities of routers
Solution Approach 1:
The routing table is segmented across multiple hierarchical domains. Instead of a single monolithic routing table that would require immense memory, each router maintains a smaller routing table containing only routes relevant to its domain and immediate neighbors. The hierarchical structure allows routers to reach distant servers through multiple hops via domain boundary routers, keeping individual routing tables within memory capacity while supporting millions of servers overall.
3Adaptability or versatility
If the number of servers increases to improve network capacity, then network scalability is improved, but hardware costs increase to unmanageable levels
Solution Approach 1:
The hierarchical domain architecture segments the network into modular units that can be independently deployed and scaled. Each domain uses standard, cost-effective routing hardware rather than requiring expensive high-capacity routers at every node. The segmentation allows incremental scaling where additional domains can be added without proportionally increasing hardware costs at existing nodes.
Solution Approach 2:
The patent introduces a hierarchical dimension to the network architecture, organizing servers into multiple levels of domains. This dimensional change allows the network to scale by adding domains at different hierarchical levels rather than uniformly increasing capacity at all nodes, thereby controlling hardware costs through selective placement of high-capacity routing resources only where needed in the hierarchy.
4Productivity
If network traffic management techniques are implemented to improve traffic engineering, then network performance is improved, but NP-complete problems become difficult to solve when the number of end points becomes large
Solution Approach 1:
Traffic engineering and load balancing computations are segmented to operate within individual domains rather than requiring global network optimization. Each domain independently performs traffic management for its local servers and traffic patterns, avoiding the NP-complete complexity of optimizing routes across the entire network of millions of servers. Domain boundary routers handle inter-domain traffic with simplified routing rules.
Data Source
AI summary
Various techniques for partitioning a computer network is disclosed herein. In certain embodiments, control plane functions (e.g., computation of network routes) and/or forwarding plane functions (e.g., routing, forwarding, switching) may be partitioned and performed individually on per domain basis based on (1) a network configuration of a particular domain (e.g., end points and/or lower-level domains in the particular domain); and (2) one or more higher-level domains connected to the particular domain in the hierarchy. Thus, a particular domain can manage various network operations of the domain without concerns regarding end points or network nodes in other domains of the hierarchy. Thus, network configuration and operation may be partitioned to reduce hardware costs and operational complexity even as the size of the overall computer networks increases.