Multicast Cloud Controller for MPLS Virtual Switches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud computing systems lack effective multicast protocol support due to the large control state requirements, which exceed host machine capacity and diminish performance, preventing efficient data delivery to groups of virtual machines (VMs) within data centers.
Innovation Solution
Implementing a Multicast Cloud Controller (MCC) that manages multicast traffic using multiprotocol label switching (MPLS) and top-of-rack switches, allowing VMs to join or leave multicast groups by updating flow table entries to forward traffic efficiently across the cloud network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multicast protocol support is implemented in cloud systems, then data delivery to groups of VMs is enabled, but control state requirements exceed host machine capacity and performance diminishes
Solution Approach 1:
The patent extracts the multicast control state management from the host machines and relocates it to the network switches. The host machine only needs to send join/leave messages, while the switch maintains the multicast group membership information and forwarding state, thereby reducing the control state burden on host machines.
Solution Approach 2:
The network switch acts as an intermediary between the host machine and the multicast group. It receives join/leave messages from the host, maintains the multicast state, and manages the distribution of multicast traffic to multiple VMs, thereby enabling multicast support without requiring the host to maintain complex control state.
2Productivity
If multicast control state is maintained on host machines, then multicast delivery is achieved, but host machine capacity is exceeded and performance diminishes
Solution Approach 1:
The patent extracts the multicast control state from the host machine and places it in the network switch. The host only needs to send simple join/leave messages, while the switch maintains the detailed multicast group membership state and forwarding information, thereby reducing the control state volume on the host machine.
Solution Approach 2:
The patent moves the multicast control state management from the computational dimension (host machine CPU and memory) to the network dimension (switch forwarding tables). This dimensional shift allows the control state to be maintained in a location optimized for network traffic management rather than general-purpose computing.
3Adaptability or versatility
If VM movement is supported in cloud systems, then flexibility is improved, but multicast state management becomes more complex
Solution Approach 1:
The network switch automatically updates its multicast state when a VM moves. When the host sends a join message for a moved VM, the switch automatically updates the multicast group membership information without requiring manual reconfiguration, thereby simplifying the management of multicast state during VM movement.
Solution Approach 2:
The system uses feedback through join/leave messages to automatically update multicast state. When a VM moves to a new host, the new host sends a join message, and the switch receives feedback about the VM's new location and updates its multicast forwarding state accordingly, enabling automatic adaptation to VM movement.
Data Source
Figure 1A~1B
Figure 2
Figure 3A~3C
AI summary
A multicast cloud controller ("MCC") in a cloud system implements a process to manage multicast traffic in a cloud network. The MCC is coupled to at least one virtualized server for hosting one or more virtual machines ("VM"), wherein the virtualized server comprises at least one virtual switch ("VS") that supports multiprotocol label switching (MPLS) and the virtual switch is coupled to a top of rack switch ("TORS") that supports MPLS. MPLS is utilized to support multicast data traffic in the cloud system such that the system and method reduces state and is scalable.