Centralized Server Load Balancing for VoIP Session Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current overload control mechanisms in VoIP and multimedia session networks are inefficient in managing load balancing across session control nodes, leading to overloading or underloading, especially in large networks with rapid traffic fluctuations, and often result in unfair resource distribution and poor customer experience.
Innovation Solution
A centralized server system that receives load status messages from session control nodes, determines the number of session set-up requests directed to each node, and selects appropriate policies to manage load balancing, ensuring efficient distribution of traffic and reducing communication complexity by allowing each node to communicate with a single central server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a centralized server system is implemented to manage load balancing across session control nodes, then load distribution optimization and prevention of overloading/underloading is achieved, but device complexity and communication infrastructure requirements increase
Solution Approach 1:
A centralized server is introduced as an intermediary component that receives load status messages from session control nodes and sends session set-up policies back to them. This mediator coordinates traffic distribution across the network, optimizing load balancing while maintaining a relatively simple implementation at the node level.
Solution Approach 2:
The system implements a feedback mechanism where session control nodes continuously report their load status to the centralized server, which then adjusts session set-up policies accordingly. This closed-loop control enables dynamic adaptation to changing network conditions, preventing both overloading and underloading of nodes.
2Reliability
If proportional discard rate is used for overload control, then some discrimination in load shedding is achieved, but it becomes impractical to match and feedback the optimal discard rate when traffic demand varies rapidly
Solution Approach 1:
The centralized server pre-calculates and maintains a set of session set-up policies that correspond to different load conditions. When a node reports its status, the server can immediately retrieve and apply the appropriate pre-prepared policy, avoiding calculation delays and enabling rapid response to traffic changes.
Solution Approach 2:
The system dynamically adjusts session set-up policies based on real-time load status feedback from nodes. The centralized server continuously monitors network conditions and modifies policy distribution accordingly, enabling the system to adapt flexibly to rapidly changing traffic demands without fixed discard rates.
3Reliability
If overly severe restriction predictions are used to protect nodes, then node protection is improved, but customer experience and network revenue are reduced due to unsuccessful sessions
Solution Approach 1:
The system uses multiple session set-up policies with different restriction parameters (e.g., varying discard rates, priority levels) rather than a single severe restriction. The centralized server selects the most appropriate policy based on current load conditions, allowing for fine-grained control that protects nodes while minimizing impact on successful session establishment.
Solution Approach 2:
Different session set-up policies are applied to different nodes or different types of traffic based on their specific load conditions and priorities. This localized approach allows severe restrictions to be applied only where and when necessary, rather than uniformly across the entire network, thereby protecting nodes while preserving customer experience where capacity is available.
4Reliability
If absolute rate restriction is signaled to neighbour nodes, then demand can be matched to available capacity and node protection is improved, but the node needs preconfigured information of other nodes and must apportion fairness amongst them
Solution Approach 1:
The centralized server consolidates the configuration and apportionment functions into a single location. Instead of each node needing preconfigured information about all other nodes and independently calculating fair apportionment, the centralized server receives load status from all nodes, determines appropriate restriction levels, and distributes policies accordingly, eliminating the need for complex distributed configuration.
Solution Approach 2:
The centralized server acts as an intermediary that manages the apportionment of traffic restrictions among nodes. It receives load status messages from all session control nodes, calculates appropriate session set-up policies considering network-wide conditions, and sends targeted policies to each node. This mediates the complexity of fair apportionment centrally rather than requiring each node to independently understand and enforce fairness rules.
Data Source
AI summary
A centralized server system (12, 18) is arranged to monitor overload restriction requirements of session control nodes (21A-D) within a globally scalable VoIP and Multimedia network (20) in order to handle session set-up requests. An overload status module (38) in the server system receives overload status messages (22) from the session control nodes and based on these messages and the demand on each session control node determined from session set-up requests, the server system (12, 18) selects which session control nodes (21A-D) as well as which of their associated interconnects to other networks to include in a session setup response (6) to a session setup request (5) from a source session control node (21A-D).