Overlay Networking Design for Substrate Service Scalability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud service providers face challenges in managing large data center footprints and resource scalability, leading to inefficiencies in cloud resource management and increased operational costs.
Innovation Solution
Implementing a new architecture that shifts services from the substrate layer to an overlay network, utilizing virtual IP addresses and dynamic routing mechanisms to optimize data center footprint and enhance scalability while maintaining core CSP services.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cloud service providers scale data centers to meet increasing cloud computing demand, then availability of cloud resources is improved, but data center footprint and operational costs increase
Solution Approach 1:
The patent segments network services into two distinct layers: substrate layer for infrastructure services and overlay layer for customer-facing services. This segmentation allows the data center to maintain a smaller footprint by moving services to the overlay network while still providing high availability through the substrate layer's distributed infrastructure.
Solution Approach 2:
The patent introduces a new dimensional layer (overlay network) above the traditional substrate network layer. By moving services to this additional dimension, the system achieves scalability and availability without proportionally increasing the physical data center footprint, as the overlay can be instantiated on existing substrate infrastructure.
2Reliability
If dedicated hosts are provided for substrate network services, then service reliability is improved, but device complexity and resource consumption increase
Solution Approach 1:
The patent implements multi-functionality by enabling overlay network nodes to perform multiple roles: they can host customer workloads, provide substrate services, and route traffic between layers. This eliminates the need for dedicated hosts for substrate services, reducing overall device complexity while maintaining service reliability through the shared infrastructure's robust design.
Solution Approach 2:
The overlay network enables services to be self-provisioned and self-managed through the overlay itself, rather than requiring separate substrate host infrastructure. The overlay's dynamic routing and service discovery mechanisms allow services to automatically locate and communicate with each other, reducing the need for dedicated management hosts.
3Ease of operation
If services are provided in the substrate layer, then connectivity with substrate services is simplified, but scalability for customer demands is reduced
Solution Approach 1:
The overlay network acts as an intermediary layer between customer workloads and substrate services. It provides simplified connectivity by abstracting the substrate infrastructure details, while simultaneously enabling scalability through its dynamic service discovery and routing capabilities. The overlay mediates between the need for simple substrate connectivity and the need for scalable customer service deployment.
Solution Approach 2:
The overlay network introduces dynamic routing and service discovery mechanisms that allow services to be dynamically added, removed, or relocated without affecting substrate connectivity. This dynamic nature enables the system to scale for customer demands while maintaining simplified connectivity to substrate services through the overlay's adaptive routing protocols.
Data Source
AI summary
A cloud service is provided and accessible via a virtual IP address in a virtual cloud network (VCN), where the VCN is part of an overlay network. A source network virtualization device (NVD) receives a first packet destined to the virtual IP address. The source NVD determined based upon a dynamic routing gateway table, the VCN mapped to the virtual IP address. Further, the source NVD identifies based upon the virtual IP address, a substrate IP address of a destination NVD that implements a virtual network interface card (VNIC) associated with the virtual IP address and communicates the first packet from the source NVD to the destination NVD using the substrate IP address.