Scalable Network Failover via Stateless Transport and State Migration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current failover solutions for decapsulation nodes in network communication channels are limited in scalability and require tight control, leading to increased complexity and costs, especially when dealing with systems using stateful transport layers or multiple decapsulation nodes.
Innovation Solution
A method is implemented to establish a reliable communication channel between an encapsulation node and a cluster of decapsulation nodes, where a primary and redundant session are initialized, and data is seamlessly switched to a redundant decapsulation node upon failure, using an EN-DN mapping table and a failure detector to manage IP addresses and maintain connectivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IP-redundancy protocols (HSRP/VRRP) are used for failover, then a single primary decapsulation node can be established, but load balancing across multiple nodes requires multiple IP addresses which complicates configuration
Solution Approach 1:
The patent segments the failover mechanism into two independent parts: (1) a stateless transport layer using simple IP redundancy protocols for basic failover, and (2) a state migration layer that handles TCP state transfer separately. This segmentation allows multiple decapsulation nodes to share a single IP address without configuration complexity while still providing load balancing and failover capabilities.
Solution Approach 2:
The patent introduces an intermediary state migration mechanism that sits between the stateless transport layer and the application layer. This intermediary handles TCP state capture, storage, and migration, allowing multiple decapsulation nodes to seamlessly take over connections without requiring complex shared state management or multiple IP addresses.
2Reliability
If FT-TCP is used to migrate TCP connections between operating system instances, then a redundant decapsulation node can be maintained in the same state as the primary server, but this requires tight coupling between nodes and a dedicated redundant node which does not scale efficiently
Solution Approach 1:
The patent separates the state management function from the transport function. Instead of requiring tight coupling through FT-TCP, it uses a stateless transport layer with an independent state migration mechanism that can operate with any decapsulation node. This allows any node in the cluster to become primary for a given IP address, enabling efficient scaling without dedicated redundant nodes.
Solution Approach 2:
The patent makes the state migration mechanism universal by designing it to work with multiple decapsulation nodes rather than requiring a dedicated redundant node. Any decapsulation node can capture and hold TCP state, and any node can take over a connection by retrieving the state from the migration mechanism, enabling flexible scaling and load distribution across the cluster.
3Adaptability or versatility
If multiple decapsulation nodes are configured with load balancing, then redundancy and scalability are improved, but existing IP-redundancy protocols do not support stateful transport layer connections
Solution Approach 1:
The patent merges two previously separate mechanisms into a unified system: (1) stateless IP redundancy protocols for basic failover and load balancing, and (2) a new state migration layer for TCP connection persistence. This combination enables multiple decapsulation nodes to provide both load balancing and stateful connection support simultaneously, resolving the limitation of existing protocols.
Solution Approach 2:
The patent introduces a state migration intermediary that bridges the gap between stateless transport protocols and stateful application requirements. This intermediary captures TCP states, stores them externally, and makes them available to any decapsulation node, enabling multi-node load balancing while maintaining connection reliability.
Data Source
AI summary
A method is provided for establishing a reliable communication channel between an encapsulation node and a cluster of decapsulation nodes across a communication network. The method is implemented at the encapsulation node and comprises the following steps. A primary session is initialized with a primary decapsulation node in the cluster of decapsulation nodes during which an assigned address is received. A redundant session is initialized with a redundant decapsulation node in the cluster of decapsulation nodes. Data is transmitted to and received from the decapsulation node until a failure is detected. Once the failure is detected, data is transmitted to and received from the redundant decapsulation node instead of the primary decapsulation node.