Multipath Failover via Dynamic Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data systems face inefficiencies in load distribution across multiple interfaces, with methods like link aggregation requiring complex configuration and not scaling well, necessitating a more effective method for protecting and accessing data.
Innovation Solution
A system and method that utilizes an intermediary to dynamically manage connections across multiple ports on a server by identifying and switching to optimal ports based on activity counters and time-outs, allowing for efficient load balancing and fault tolerance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If link aggregation is used to spread load across multiple interfaces, then load distribution is achieved, but configuration complexity increases and scalability deteriorates
Solution Approach 1:
The system enables automatic load distribution across multiple network interfaces without requiring manual configuration. The load balancer automatically detects available interfaces, assigns weights, and distributes traffic based on configured policies, eliminating the need for complex manual setup while maintaining efficient load distribution
Solution Approach 2:
The load balancing mechanism is designed to work with multiple types of network interfaces (Ethernet, wireless, virtual) and various distribution algorithms (round-robin, weighted, least-connections) through a unified interface, allowing the same system to handle different interface types and scaling scenarios without requiring separate configuration approaches
2Productivity
If multiple interfaces are utilized for load spreading, then bandwidth increases, but system reliability against interface failures decreases
Solution Approach 1:
The load balancer continuously monitors the health status of each network interface through active probing and response time measurement. When an interface fails or degrades, the system automatically detects the failure, removes the problematic interface from the active pool, and redistributes its traffic load to healthy interfaces, thereby maintaining both high bandwidth utilization and system reliability
Solution Approach 2:
The system pre-configures backup interfaces and establishes failover policies before failures occur. When primary interfaces are available, traffic is distributed across all interfaces to maximize bandwidth. Upon failure detection, the pre-established backup interfaces automatically activate to maintain connectivity and redistribute load, ensuring continuous operation without service interruption
3Reliability
If manual interface configuration is performed for failover protection, then reliability improves, but operation complexity increases
Solution Approach 1:
The load balancing system automatically performs interface health monitoring, failure detection, and failover execution without requiring manual intervention. Administrators only need to define high-level policies (which interfaces to use, distribution algorithm preferences), and the system autonomously manages the complex failover logic, interface monitoring, and traffic redistribution, thereby achieving high reliability with minimal operational complexity
Solution Approach 2:
The system pre-configures failover policies, interface priorities, and backup routes in advance through simple administrative input. When failures occur, the pre-established configurations are automatically executed, eliminating the need for operators to manually configure complex failover scenarios during critical failure events. This preliminary setup phase is straightforward and does not require deep technical expertise
Data Source
AI summary
A method, article of manufacture, and apparatus for processing data. In some embodiments, this includes detecting a faulty interface, removing outstanding jobs from the faulty interface, distributing the outstanding jobs to other interfaces. In some embodiments, distributing the outstanding jobs includes distributing a first outstanding job to a first lowest counter interface. In some embodiments, a second outstanding job is distributed to a second lowest counter interface.