Virtual Network Management via Computation Domains
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data center network virtualization, heterogeneous hypervisors managed by different SDN controllers cannot directly communicate via overlay networks due to the inability of different SDN controllers to sense each other's networks, leading to complex and inefficient packet forwarding.
Innovation Solution
Creating computation domains for managing heterogeneous hypervisors and using topology management information to enable unified control by SDN controllers, allowing VMs across different hypervisors to communicate through overlay networks by distributing configuration information to virtual switches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different SDN controllers are used to manage heterogeneous hypervisors, then each hypervisor can be managed independently, but direct communication between VMs on different hypervisors becomes unavailable
Solution Approach 1:
The patent introduces a computation domain as an intermediary layer between different SDN controllers and heterogeneous hypervisors. Each SDN controller manages its own hypervisors through computation domains, which act as mediators to enable communication between VMs on different hypervisors. The computation domain includes topology management information that facilitates this intermediary communication, resolving the contradiction by allowing both independent management and reliable intercommunication.
2Adaptability or versatility
If multiple SDN controllers manage heterogeneous hypervisors, then management flexibility is improved, but packet forwarding complexity increases
Solution Approach 1:
The patent segments the network management into computation domains, where each SDN controller manages a specific domain containing topology management information. This segmentation allows flexible management of heterogeneous hypervisors while simplifying packet forwarding within each domain. The topology management information within each computation domain provides the necessary routing details, reducing overall system complexity despite the multi-controller architecture.
3Adaptability or versatility
If different hypervisor managers are used, then system heterogeneity is supported, but network management complexity increases
Solution Approach 1:
The patent creates a universal computation domain structure that can accommodate multiple types of hypervisor managers. Each computation domain is designed with multi-functionality to handle different hypervisor types while maintaining a consistent management framework. The topology management information within each domain provides universal routing capabilities, reducing network management complexity despite supporting heterogeneous hypervisors.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In an example, a method for managing a virtual network is provided. According to the method, first and second domains are respectively created for heterogeneous first and second hypervisor managers. The first hypervisor manager manages a first virtual machine (VM) and first virtual switch (vSwitch). The second hypervisor manager manages a second VM and second vSwitch. Based on topology of the virtual network, topology management information may be created, which comprises first and second virtual network topology information respectively for the first and second domains, and first and second configuration information respectively for the first and second vSwitches. The topology management information may be distributed to the first and second hypervisor managers after being associated with the first and second domains, so the first and second vSwitches implement packet forwarding between the first and second VMs according to the first and second configuration information respectively.