Network Interface Card Overlay Routing for Virtual Machine Traffic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Virtual networks in distributed overlay networks are constrained by physical network topologies, limiting the flexibility and independence of data packet routing between virtual machines.
Innovation Solution
A data traffic module on a network interface card identifies physical path translations independent of physical network topology, encapsulates data packets with these translations, and sends them to destination virtual machines over the physical network, bypassing hypervisor dependencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual networks are constrained by physical network topology, then routing follows physical infrastructure, but flexibility and independence of data packet routing between virtual machines deteriorates
Solution Approach 1:
The patent segments the network routing function into two independent layers: physical network topology (infrastructure layer) and virtual network routing (logical layer). By encapsulating data packets with overlay network headers containing virtual machine identifiers, the system separates virtual routing decisions from physical path constraints, allowing virtual networks to be segmented and managed independently from the underlying physical infrastructure.
Solution Approach 2:
The patent introduces an intermediary encapsulation mechanism (overlay network header) that mediates between virtual machine communication requirements and physical network routing. This intermediary layer translates virtual network identifiers into physical routing information, enabling flexible virtual routing while relying on the stable physical infrastructure without direct coupling.
2Ease of operation
If virtual switch is tightly coupled to hypervisor platform dependencies, then host computer system management is simplified, but independence from physical topology constraints deteriorates
Solution Approach 1:
The patent adds another dimension to virtual switch operation by implementing overlay network routing at the data packet level rather than relying solely on hypervisor-level virtual switch configuration. This dimensional shift allows the system to maintain simple hypervisor management while achieving topology independence through encapsulation-based routing decisions that operate independently of hypervisor platform dependencies.
Solution Approach 2:
The patent extracts the topology-independent routing logic from the hypervisor-coupled virtual switch into a separate encapsulation mechanism. By taking out the routing intelligence into overlay network headers that travel with data packets, the system maintains ease of virtual switch management while achieving independence from both hypervisor platform dependencies and physical topology constraints.
3Reliability
If data packets are routed according to physical network topology, then router placement constraints are followed, but network efficiency and flexibility for concurrent virtual networks deteriorates
Solution Approach 1:
The patent segments network traffic into individual encapsulated packets, each carrying its own virtual machine identifier and routing information. This segmentation allows each packet to be independently routed based on virtual network requirements rather than physical topology, enabling multiple virtual networks to operate concurrently with optimized paths while the underlying physical network maintains stable router placements.
Solution Approach 2:
The patent introduces dynamic routing capabilities at the packet level through overlay network encapsulation. While physical router placements remain static and stable, the virtual routing paths become dynamic and adaptable, allowing network efficiency to be optimized for each virtual machine communication pair without compromising the stability of the physical infrastructure.
Data Source
AI summary
An approach is provided in which a data traffic module executing on a network interface card receives a data packet initiated by a first virtual machine with a destination at a second virtual machine. The data traffic module identifies one or more physical path translations corresponding to a logical connectivity that is independent of physical topology constraints of a physical network. In turn, the data traffic module encapsulates the data packet with the one or more physical path translations and sends the encapsulated data packet to the second virtual machine over the physical network.