VM Tracking via Augmented VDP Protocol in Data Center Fabric
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data center orchestration engines lack visibility into the physical network infrastructure, making it difficult to track virtual machines (VMs) and their physical attachments within the data center fabric, leading to inefficient network management and troubleshooting.
Innovation Solution
Implementing an augmented Virtual Station Interface Discovery and Configuration Protocol (VDP) that includes Type-Length-Value (TLV) elements to convey VM identities from virtual switches to physical switches, enabling the data center network manager to query and track VM locations within the fabric.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If ARP and MAC tables are dumped to track VM locations, then some tracking information is obtained, but the process becomes unwieldy and difficult to correlate across databases
Solution Approach 1:
The patent introduces an intermediary component (VM tracking system with VDP protocol) that mediates between the virtualized infrastructure and physical network fabric. This intermediary translates VM identity information into a standardized format that can be efficiently tracked across the network fabric, eliminating the need for administrators to manually correlate ARP and MAC tables from multiple switches.
Solution Approach 2:
The patent segments the tracking function into distinct components: VDP protocol for information exchange, VSI TLV elements for structured data representation, and distributed storage at network elements. This segmentation allows each component to handle specific aspects of VM tracking independently, reducing overall system complexity.
2Measurement precision
If per-interface or per-vNIC tracking is implemented, then detailed network information is captured, but it becomes difficult to correlate information to per end host/VM basis
Solution Approach 1:
The patent merges per-interface and per-VM tracking by incorporating VM identity information directly into the VDP protocol exchanges. This allows the system to maintain detailed per-interface tracking data while simultaneously correlating it to the VM level through the VSI TLV elements, providing both granular and aggregated views.
Solution Approach 2:
The VDP protocol is enhanced to serve multiple functions simultaneously: it maintains backward compatibility for basic virtual switch discovery while also carrying VM identity information through VSI TLV elements. This universal approach allows a single protocol to handle both traditional switching operations and advanced VM tracking requirements.
3Loss of information
If orchestration engines are given visibility into the fabric, then VM tracking capability is improved, but the system complexity and integration requirements increase
Solution Approach 1:
The patent positions the enhanced VDP protocol and VSI TLV elements as an intermediary layer between orchestration engines and the network fabric. This intermediary provides the necessary visibility and tracking information without requiring deep integration or modification of the orchestration engine itself, maintaining a clean architectural boundary.
4Loss of information
If switches relay information about VM physical attachments, then network administrator visibility is improved, but the switch processing overhead and information management complexity increase
Solution Approach 1:
The patent changes the parameter representation by introducing VSI TLV elements that encode VM identity information in a standardized, compact format. This parameter transformation allows switches to efficiently store, process, and relay VM attachment information without significant increases in memory or processing requirements.
Data Source
AI summary
A method is provided in one example embodiment and includes receiving at a first network element a packet from a second network element; processing the packet at the first network element to obtain information regarding an identity of a virtual machine (“VM”) hosted by the second network element contained within the packet; and storing at the first network element the identifying information. The identifying information stored at the first network element is accessible by at least one third network element. In some embodiments, the first network element comprises a physical switch and the second network element comprises a virtual switch.