RBridge Host Route Distribution for VM Mobility Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In network environments with virtual machines hosted across multiple data centers, existing technologies face challenges in efficiently distributing host route information to routing bridges (RBridges) to ensure seamless communication between virtual machines, particularly when virtual machines move between data centers, leading to network traffic disruptions and scalability issues.
Innovation Solution
The introduction of new BGP attributes such as nickname, mobility, and forwarding information base (FIB) attributes allows RBridges to intelligently distribute and install host route information, determining the next hop and managing hardware FIB overflow, thereby optimizing communication paths and reducing network latency and bandwidth usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual machines are hosted by physical servers in different data centers and communicate indirectly through border RBridges, then routing information can be distributed across data centers, but network latency increases due to additional routing hops and lookups
Solution Approach 1:
The patent pre-computes and caches routing information (host routes) in RBridges before actual virtual machine communication occurs. When a virtual machine moves between data centers, the routing information is proactively updated and distributed to relevant RBridges in advance, so that when communication occurs, the routing decisions are already cached and do not require real-time computation or multiple lookups, thereby reducing network latency while maintaining reliable routing distribution
2Adaptability or versatility
If host route information is distributed to all RBridges in the network, then virtual machine mobility is supported, but device complexity and scalability are reduced due to increased routing table sizes
Solution Approach 1:
The patent implements selective host route distribution where each RBridge receives and stores only the routing information relevant to its location and function in the network. Instead of distributing all host routes to all RBridges, the system determines which RBridges need which routing information based on their position in the network topology and their role in virtual machine communication paths. This localized approach maintains virtual machine mobility support while significantly reducing routing table sizes and device complexity
Solution Approach 2:
The patent segments the routing information distribution by data center and by RBridge function. Host routes are divided into local data center routes and remote data center routes, and further segmented by which RBridges need to cache them. This segmentation allows the network to scale by adding more RBridges and data centers without proportionally increasing the routing table size at each device, as each RBridge only maintains the segmented portion of routing information relevant to its segment
3Reliability
If routing messages are distributed to all RBridges for host route information, then comprehensive routing coverage is achieved, but network bandwidth is consumed by redundant routing message transmissions
Solution Approach 1:
The patent implements a feedback mechanism where RBridges send acknowledgments to the originating RBridge when they receive host route information. The originating RBridge tracks which RBridges have received routing messages and uses this feedback information to determine whether to transmit the same routing message again. This feedback-based approach ensures comprehensive routing coverage across the network while avoiding redundant transmissions, as the originating RBridge only sends routing information to RBridges that have not yet received it, thereby conserving network bandwidth
Data Source
AI summary
Techniques are presented for distributing host route information of virtual machines to routing bridges (RBridges). A first RBridge receives a routing message that is associated with a virtual machine and is sent by a second RBridge. The routing message comprises of mobility attribute information associated with a mobility characteristic of the virtual machine obtained from an egress RBridge that distributes the routing message. The first RBridge adds a forwarding table attribute to the routing message that indicates whether or not the first RBridge has host route information associated with the virtual machine in a forwarding table of the first RBridge. The first RBridge also distributes the routing message including the mobility attribute information and the forwarding table attribute, to one or more RBridges in the network.