Smart-vLB Hierarchical Load Balancer for Probe Capacity Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern communication networks face challenges in real-time monitoring and troubleshooting to maintain performance and quality of service due to their complexity and the increasing dominance of virtual network entities, which require efficient load balancing and throttling mechanisms to manage data distribution effectively.
Innovation Solution
A hierarchical load balancing system, referred to as Smart-vLB, comprising non-stateful and stateful virtual load balancers that distribute and throttle packets based on contextual and non-contextual attributes, such as IMSI or APN, to optimize data processing and prevent probe capacity oversubscription, while ensuring session integrity and dynamic adjustment to network traffic.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a hierarchical load balancing system with stateful and non-stateful virtual load balancers is implemented, then network traffic distribution and probe capacity management are improved, but system complexity increases
Solution Approach 1:
The load balancing system is segmented into multiple hierarchical levels: non-stateful virtual load balancers (NS-vLBs) at the first level for basic packet distribution, and stateful virtual load balancers (S-vLBs) at the second level for session-aware routing. This segmentation allows each component to specialize in specific functions, improving overall traffic distribution efficiency while managing complexity through modular design
Solution Approach 2:
The stateful virtual load balancer acts as an intermediary between the non-stateful load balancers and the network probes. It receives session information from NS-vLBs, maintains session state, and makes intelligent routing decisions based on probe capacity and session affinity requirements, thereby managing the complexity of stateful operations in a distributed system
2Reliability
If session affinity routing is implemented to ensure packets from the same session are processed by the same probe, then session integrity is improved, but load distribution across probes deteriorates
Solution Approach 1:
The load balancing system dynamically adjusts session affinity routing based on real-time probe capacity and load conditions. The S-vLB monitors probe status and can dynamically change routing decisions, allowing session packets to be redirected to different probes when capacity constraints change, thereby maintaining both session integrity and efficient load distribution
Solution Approach 2:
The system changes routing parameters based on session type and probe capacity. For sessions with strict affinity requirements, the system maintains consistent probe assignment, while for other sessions, it allows more flexible distribution. This parameter-based approach enables the system to optimize the trade-off between session integrity and load distribution efficiency
3Reliability
If throttling mechanisms are implemented to prevent probe capacity oversubscription, then quality of service is improved, but packet processing throughput deteriorates
Solution Approach 1:
The load balancers perform preliminary assessment of probe capacity and session requirements before routing packets. The S-vLB pre-evaluates whether a probe has sufficient capacity to handle additional session traffic, and only routes packets to probes that can maintain QoS standards, thereby preventing oversubscription while maximizing throughput
Solution Approach 2:
The system implements feedback mechanisms where the S-vLB continuously monitors probe capacity and performance metrics. Based on this feedback, it dynamically adjusts throttling decisions, allowing higher throughput when probes have excess capacity and applying stricter throttling when capacity is constrained, thus optimizing both QoS and throughput
Data Source
AI summary
A method of processing packets propagated over a packet switched communications network having a control plane, user plane, and a plurality of probes, the method comprising: receiving at least one control plane packet associated with creating at least one user session in the network; selecting a set of user sessions from the at least one user session; determining at least one target feature that characterizes packets propagated over the network; and load balancing all packets sharing the at least one target feature that belong to a same user session of the set of user sessions to a same probe of the plurality of probes for processing by the probe.