Access Switch ARP Proxying for VM Migration Flooding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In data centers with virtual machines (VMs), VM migration across different Top of Rack (TOR) switches leads to exponential ARP flooding and a significant increase in MAC address table size due to scattered subnets and disjointed addresses, causing inefficiencies in packet forwarding and address management.

Innovation Solution

The method involves an access switch receiving a Layer 3 packet with a Virtual Network ID (VNID) and determining a Designated Node (DN) to generate a Layer 2 frame with the DN's MAC address, and TOR switches handling migrated VMs by determining the new TOR switch for packet forwarding and updating MAC tables to reduce ARP flooding and table size.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If VM migration is enabled across TOR switches in a flat Layer 2 network, then VM mobility and flexibility are improved, but ARP flooding and MAC table size increase exponentially

Engineering Contradiction:
ImproveVM mobilityVSAvoidARP table entries
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent introduces an access switch as an intermediary between VMs and TOR switches. This access switch performs ARP proxying and MAC address translation, acting as a mediator that prevents ARP broadcasts from propagating across the entire network. The access switch maintains local ARP tables while translating MAC addresses for cross-TOR communication, thereby containing ARP flooding to local segments only.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The network is segmented into multiple access switch domains, each maintaining its own ARP table. By dividing the flat Layer 2 network into smaller broadcast domains through access switches, ARP flooding is contained within each segment rather than propagating across all TOR switches. This segmentation reduces the global ARP table size while maintaining VM mobility across segments.

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If VLAN spans across multiple racks to enable VM migration, then VM mobility is improved, but ARP broadcast flooding increases

Engineering Contradiction:
ImproveVM migration capabilityVSAvoidARP broadcast flooding
Core Design Contradiction:
Adaptability or versatilityVSObject-generated harmful factors

Solution Approach 1:

The access switch serves as an intermediary that intercepts and processes ARP broadcasts before they propagate across VLANs spanning multiple racks. By performing ARP proxying at the access switch level, the patent prevents ARP broadcasts from flooding across rack boundaries while still enabling VM migration through MAC address translation and forwarding.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If Access switch floods packets over the whole Layer 2 Network for migrated VMs, then packet delivery is ensured, but network efficiency deteriorates

Engineering Contradiction:
Improvepacket deliveryVSAvoidnetwork efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The access switch implements feedback mechanisms by maintaining MAC address tables that track VM locations across TOR switches. When a packet needs to be forwarded to a migrated VM, the access switch queries its MAC table to determine the destination TOR switch, enabling directed forwarding instead of blind flooding. This feedback-driven approach ensures reliable packet delivery while improving network efficiency.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The access switch acts as an intermediary that performs intelligent packet forwarding based on MAC address translation and TOR switch mapping. Instead of flooding packets across the entire Layer 2 network, the access switch translates destination MAC addresses and forwards packets only to the appropriate TOR switch, ensuring reliable delivery while eliminating unnecessary network traffic.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Ease of operation

If Layer 3 capabilities are enabled on TOR switch to manage disjointed subnets, then subnet management is improved, but system complexity and cost increase

Engineering Contradiction:
Improvesubnet managementVSAvoidTOR switch complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The patent introduces an access switch as an intermediary that handles subnet management and MAC address translation functions, thereby relieving the TOR switch of Layer 3 management complexity. The access switch maintains subnet-to-MAC mappings and performs intelligent forwarding, enabling disjointed subnet management without requiring TOR switches to implement complex Layer 3 capabilities.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent extracts subnet management and ARP management functions from the TOR switch and relocates them to the access switch. By taking out these complex Layer 3 management capabilities from the TOR switch, the patent simplifies the TOR switch while maintaining the ability to manage disjointed subnets through the access switch's MAC address translation and forwarding mechanisms.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS9270590B2Method, apparatus and system for transmitting packets in virtual network with respect to a virtual machine (VM) migration
Publication Date: 2016.02.23 HUAWEI TECH CO LTD
  • US9270590B2 patent drawing
  • US9270590B2 patent drawing
  • US9270590B2 patent drawing

AI summary

The embodiments of the present disclosure provide a method for transmitting packet in a Virtual Network. In the method, an access switch receives a Layer 3 packet carrying a VNID (Virtual Network IDentifier) from a VM in a remote Data Center. The access switch determines a DN (Designated Node) corresponding to the VNID and generates a Layer 2 frame according to the Layer 3 packet, where the Layer 2 frame includes the MAC (Media Access Control) address of the DN. The access switch to the DN transmits the Layer 2 frame according to the MAC address of the DN such that the DN determines a Layer 3 destination address according to the Layer 2 frame. This avoids packet flooding in Data Center when VM was migrated.