Fused Data Center Fabric Network Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches to scaling networks in data centers and IP-based environments are costly and complex due to the need for full connectivity and 1:1 oversubscription, leading to high overhead and inefficiency, especially when using large, expensive hardware.
Innovation Solution
Implementing a network design that uses groups of network switches with reduced connectivity between tiers, introducing oversubscription between spine and egress switches, and fusing edges of deployment units to reduce the number of devices needed, allowing for scalable and efficient host connections without increasing between-tier connections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional routing protocols and full connectivity are used to scale networks, then network capacity and reliability are improved, but device complexity and installation complexity increase significantly
Solution Approach 1:
The network is divided into multiple fabrics that can be independently managed and scaled. Each fabric operates as a separate unit with its own routing domain, allowing incremental deployment and reduced installation complexity while maintaining overall network capacity and reliability through fabric interconnection.
Solution Approach 2:
Border gateway protocols serve as intermediaries between fabrics, enabling communication and routing between separate network domains without requiring full connectivity across the entire network. This mediator layer simplifies the routing complexity by confining protocol processing to fabric boundaries.
2Productivity
If large, expensive hardware is used to provide sufficient network capacity, then network performance is improved, but cost increases significantly
Solution Approach 1:
The network capacity is distributed across multiple smaller, less expensive hardware devices organized into fabrics. This segmentation allows the use of cost-effective equipment while achieving the required overall network capacity through aggregation of multiple fabric resources.
Solution Approach 2:
Multiple fabrics can be created as copies or replicas of a base fabric design, allowing standardized, cost-effective deployment patterns to be replicated across the network. This copying approach reduces design costs and enables efficient resource utilization across multiple fabric instances.
3Adaptability or versatility
If the number of devices is increased to handle conventional routing protocols, then network scalability is improved, but overhead and expenditure increase
Solution Approach 1:
Routing overhead is segmented and confined to fabric boundaries rather than requiring processing across the entire network. Border gateway protocols handle inter-fabric routing with reduced complexity compared to conventional protocols, allowing scalability without proportional increases in overhead.
Solution Approach 2:
Conventional routing protocols are applied partially only at fabric boundaries rather than throughout the entire network. This partial application reduces the overall routing overhead while maintaining scalability, as most internal fabric communication uses simplified routing mechanisms.
Data Source
AI summary
Efficient and highly-scalable network solutions are provided that each utilize deployment units based on Clos networks, but in an environment such as a data center of Internet Protocol-based network. Each of the deployment units can include multiple stages of devices, where connections between devices are only made between stages and the deployment units are highly connected. In some embodiments, the level of connectivity between two stages can be reduced, providing available connections to add edge switches and additional host connections while keeping the same number of between-tier connections. In some embodiments, where deployment units (or other network groups) can be used at different levels to connect other deployment units, the edges of the deployment units can be fused to reduce the number of devices per host connection.