Service Chain Packet Processing via Hypervisor Datapath Bypass

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional packet processing in Software-Defined Networking (SDN) environments using service chains is inefficient due to the need for packets to be transported in and out of virtual switches multiple times, leading to bottlenecks and scalability issues, especially with a large number of virtual machines in the service chain.

Innovation Solution

Packets are transported through established datapaths directly between virtual machines in the service chain, bypassing the virtual switch, thereby reducing intervention by the hypervisor and minimizing the risk of virtual switch bottlenecks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If packets are transported through virtual switches multiple times for service chain processing, then service chain functionality is achieved, but packet processing efficiency deteriorates due to bottlenecks

Engineering Contradiction:
Improvepacket processing efficiencyVSAvoidvirtual switch intervention
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts the packet forwarding function from the virtual switch and implements it directly in the hypervisor. The hypervisor now handles packet routing between VMs in the service chain without requiring packets to enter and exit the virtual switch multiple times, thereby eliminating the virtual switch bottleneck while maintaining service chain functionality.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a new intermediary mechanism where the hypervisor acts as the mediator between VMs in the service chain. Instead of packets repeatedly passing through the virtual switch, the hypervisor directly routes packets between the appropriate VMs, reducing unnecessary hops and improving processing efficiency.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If conventional virtual switch-based packet processing is used, then service chain deployment is simple, but scalability deteriorates with large numbers of virtual machines

Engineering Contradiction:
Improveservice chain scalabilityVSAvoidperformance with large number of VMs
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent segments the packet processing function from the virtual switch layer and assigns it to the hypervisor layer. This segmentation allows the system to handle large numbers of VMs more efficiently, as the hypervisor can directly manage packet routing without the virtual switch becoming a bottleneck, thereby improving scalability.

Inventive Principle:
Principle #1Segmentation

3Productivity

If packets pass through virtual switch multiple times, then service chain processing is achieved, but processing time increases due to repeated intervention

Engineering Contradiction:
Improvepacket processing speedVSAvoidprocessing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by having the hypervisor pre-establish direct packet routing paths between VMs in the service chain. This eliminates the need for packets to repeatedly enter and exit the virtual switch during processing, reducing processing time while maintaining the required service chain functionality.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS10581730B2Packet processing using service chains
Publication Date: 2020.03.03 VMWARE INC
  • US10581730B2 patent drawing
  • US10581730B2 patent drawing
  • US10581730B2 patent drawing

AI summary

Example methods are provided for a host to perform packet processing using a service chain in a software-defined networking (SDN) environment. The method may comprise establishing a datapath between a first virtualized computing instance and a second virtualized computing instance in the service chain, the datapath bypassing a forwarding element to which the first virtualized computing instance and the second virtualized computing instance are connected. The method may also comprise: the first virtualized computing instance obtaining a packet that requires processing by the service chain, and performing packet processing according to a first service. The method may further comprise: the second virtualized computing instance obtaining the processed packet via the datapath, and performing packet processing according to a second service.