Distributed Routing Domains for Multi-Subnet VM Migration

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing virtual datacenters face challenges in deploying and managing virtual machines (VMs) across multiple subnets due to the need for IP address changes, which disrupt network, security, and compliance policies, and require cumbersome gateway setups, limiting user adoption of cloud services.

Innovation Solution

A distributed routing domain is implemented within a network-virtualized datacenter, allowing tenants to maintain their own IP addresses and network topology through a virtualization module that enforces a multi-subnet routing topology without standalone physical routers, using Generic Routing Encapsulation (GRE) for packet encapsulation and routing between VMs on different host machines.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a single virtual subnet is provided for VMs, then network setup is simplified, but VMs in different subnets cannot communicate directly and must go through physical gateways, introducing overhead and complexity

Engineering Contradiction:
Improvenetwork setupVSAvoidgateway routing complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent merges multiple virtual subnets into a single distributed routing domain, allowing VMs from different subnets to communicate directly without traversing external physical gateways. The virtual network fabric combines multiple subnets while maintaining their logical separation, eliminating the need for complex gateway routing configurations.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces a virtual network overlay as an intermediary layer between VMs in different subnets. This virtual overlay uses encapsulation (such as GRE or VXLAN) to carry traffic between subnets, replacing the need for physical gateways and simplifying network setup while maintaining subnet isolation.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of manufacture

If IP addresses are changed when moving VMs to cloud environment, then cloud provider network management is simplified, but existing network, security, and compliance policies must be rewritten, increasing deployment complexity

Engineering Contradiction:
Improvecloud deploymentVSAvoidpolicy management complexity
Core Design Contradiction:
Ease of manufactureVSDevice complexity

Solution Approach 1:

The patent segments the network identity into two parts: the tenant's logical IP addressing scheme and the cloud provider's physical network addressing. This segmentation allows VMs to retain their original tenant IP addresses while the cloud provider manages physical addressing separately, eliminating the need to rewrite security and compliance policies.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates a virtual copy of the tenant's network environment in the cloud, where VMs are assigned their original IP addresses from the tenant's address space. This virtual copy maintains the same network identity as the on-premises environment, allowing seamless migration without policy changes while the physical infrastructure uses different addressing.

Inventive Principle:
Principle #26Copying

3Reliability

If standalone physical routers or VM routers are used to enforce multi-subnet routing topology, then routing between subnets is achieved, but device complexity and deployment overhead increase

Engineering Contradiction:
Improverouting functionalityVSAvoidrouter infrastructure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the routing functionality from standalone physical routers or VM routers and integrates it directly into the hypervisor and virtual network stack. This extraction eliminates the need for separate router infrastructure while maintaining multi-subnet routing capabilities through virtual switching and encapsulation mechanisms.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent makes the virtual switch multi-functional by enabling it to perform both switching and routing functions. The virtual switch can forward traffic within subnets using standard switching and perform inter-subnet routing using encapsulation, eliminating the need for separate router devices and simplifying the infrastructure.

Inventive Principle:
Principle #6Universality (Multi-functionality)

4Productivity

If VMs are migrated to different physical machines, then resource utilization is improved, but network connectivity and IP address consistency must be maintained, increasing management complexity

Engineering Contradiction:
Improveresource utilizationVSAvoidnetwork management complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent implements dynamic IP address assignment at the virtual network interface level, allowing VMs to maintain their logical IP addresses while their physical location changes. The virtual network stack dynamically updates the mapping between logical IP addresses and physical host locations, enabling seamless migration without requiring IP address changes or complex manual reconfiguration.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS9042384B2Distributed routing domains in multi-tenant datacenter virtual networks
Publication Date: 2015.05.26 MICROSOFT TECHNOLOGY LICENSING LLC
  • US9042384B2 patent drawing
  • US9042384B2 patent drawing
  • US9042384B2 patent drawing

AI summary

A distributed routing domain is disclosed wherein each user or tenant can deploy a multi-subnet routing topology in a network-virtualized datacenter. A virtualization module implements the distributed routing domain and enforces a multi-subnet routing topology in a distributed fashion without requiring a standalone physical router or VM router. The topology and the routing rules are distributed in a network virtualization module on each hypervisor host, and collectively realize the multi-subnet topology for a virtual network over any physical network topology.