Persistent Load Balancer Routing via SDN Controller
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multiple persistent load balancer systems face challenges in maintaining session state consistency across front-end servers, leading to routing errors and scalability issues when the routing sub-network changes, requiring either single persistent servers, state synchronization, or client-provided state information, which introduce latency and hinder system scaling.
Innovation Solution
A system with a Software Defined Networking (SDN) controller that distributes a common routing algorithm and destination distribution list to network nodes and persistent load balancer devices, allowing them to determine the correct front-end server for data flows and redirect them as needed, eliminating the need for state synchronization and enhancing scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple persistent load balancer devices are used, then system scalability and performance are improved, but routing consistency and session state reliability deteriorate when the routing sub-network changes
Solution Approach 1:
The patent introduces a controller as an intermediary that distributes a common routing algorithm and destination distribution list to all persistent load balancer devices. This mediator ensures that all devices use consistent routing logic without needing to synchronize session state, resolving the contradiction between having multiple scalable load balancers and maintaining routing consistency.
Solution Approach 2:
The patent changes the routing parameter from device-specific session state to a common routing algorithm with destination distribution list. By using five-tuple hashing and a shared routing strategy distributed by the controller, the system achieves both scalability (multiple load balancers) and consistency (uniform routing decisions) without requiring state synchronization.
2Reliability
If state information is synchronized between front end servers, then session state consistency is improved, but system latency and complexity increase
Solution Approach 1:
The patent extracts the session state storage function from the persistent load balancer devices themselves and centralizes it in the controller's destination distribution list. This allows load balancer devices to make routing decisions based on the common algorithm without accessing or synchronizing actual session state, eliminating synchronization latency while maintaining consistency.
Solution Approach 2:
The controller acts as an intermediary that provides the destination distribution list to all load balancer devices. Instead of devices synchronizing state directly with each other (which causes latency), they all query the centralized controller using a common algorithm, achieving consistency without the performance penalty of peer-to-peer synchronization.
3Productivity
If conventional ECMP routing with five tuple hashing is used, then load distribution is improved, but session persistence reliability deteriorates when routing changes occur
Solution Approach 1:
The patent applies preliminary action by having the controller distribute the common routing algorithm and destination distribution list to all persistent load balancer devices before routing decisions are made. This pre-distribution ensures that all devices have consistent routing logic in place, so when routing changes occur, they all adapt uniformly, maintaining session persistence reliability while preserving load distribution benefits.
Data Source
AI summary
A multiple persistent load balancer system includes a routing sub-network having a plurality of routing sub-network nodes that are coupled to a plurality of load balancer devices. A controller provides a common routing algorithm and a destination distribution list to each of the routing sub-network nodes and the load balancer devices. A first persistent load balancer device receives a first data flow from at least one of the routing sub-network nodes, and uses the common routing algorithm and the destination distribution list to determine whether the first data flow should be redirected to a second persistent load balancer device. If not, the first persistent load balancer device processes the first data flow to retrieve a resource from a server. If so, the first persistent load balancer device forwards the first data flow through at least one of the sub-network nodes to the second persistent load balancer device.