Performance Load Balancer Segmentation for Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The rapid deployment of 5G networks and diverse end devices with different air interface types poses challenges for network optimization, as existing solutions often compromise between latency and throughput, leading to suboptimal performance for various application services and devices.
Innovation Solution
A carrier-level performance optimization solution is implemented using a performance-oriented load balancer in the core network, which partitions servers into smaller pools with customized optimization profiles and distributes data packet traffic based on congestion control algorithms like TCP BBR or TCP CUBIC, ensuring optimal performance without modifying client- or server-side software.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single network optimization profile is applied to all end devices, then device complexity is reduced, but network performance and quality of experience deteriorate for diverse device types and application services
Solution Approach 1:
The patent segments the network optimization by dividing end devices into different groups based on air interface types (e.g., 5G NR, LTE) and application service categories. Each segment receives a customized optimization profile tailored to its specific requirements, resolving the contradiction between simplified management and optimized performance for diverse devices.
Solution Approach 2:
The patent applies local quality by assigning different optimization profiles to different device groups and application services. Each group receives locally optimized parameters (e.g., throughput-oriented profiles for video streaming, latency-oriented profiles for gaming) rather than a uniform configuration, thereby improving overall network performance while maintaining manageable complexity through systematic categorization.
2Speed
If a throughput-oriented network optimization profile is applied, then data throughput is improved, but latency and latency jitter increase leading to degraded quality of experience for time-sensitive applications
Solution Approach 1:
The patent implements dynamic optimization profiles that can be selectively applied based on the application service requirements. Time-sensitive applications receive latency-oriented profiles with dynamic parameter adjustments (e.g., lower buffer sizes, aggressive congestion control) while throughput-intensive applications receive throughput-oriented profiles, allowing the system to adaptively balance throughput and latency based on real-time needs.
Solution Approach 2:
The patent changes network parameters dynamically based on application service type. For time-sensitive applications, parameters such as buffer sizes, congestion control thresholds, and scheduling priorities are adjusted to minimize latency, while for throughput-intensive applications, parameters are optimized for maximum data rate, thereby resolving the throughput-latency trade-off.
3Loss of time
If a latency-oriented network optimization profile is applied, then latency is reduced, but data throughput deteriorates for throughput-intensive application services
Solution Approach 1:
The system dynamically selects between latency-oriented and throughput-oriented profiles based on application service characteristics. Throughput-intensive applications (e.g., video streaming, file downloads) are automatically assigned throughput-optimized profiles that prioritize data rate, while time-sensitive applications receive latency-optimized profiles, ensuring each application operates at its performance optimum without compromising the other.
Solution Approach 2:
The patent applies parameter changes specific to throughput-intensive applications by adjusting network parameters such as increasing buffer sizes, modifying congestion control algorithms, and optimizing scheduling policies to maximize throughput while maintaining acceptable latency levels for these specific application types.
4Speed
If 5G end devices with mmWave capability use a network optimization profile designed for their high throughput, then their performance is maximized, but LTE end devices experience buffer bloat and packet drops due to overwhelming data volume
Solution Approach 1:
The patent segments the device population by air interface type (5G NR vs. LTE) and applies device-specific optimization profiles. 5G devices with mmWave capability receive profiles optimized for high throughput with appropriate buffer management, while LTE devices receive profiles with conservative data rate expectations and enhanced packet handling, preventing buffer bloat and packet drops by matching optimization parameters to actual device capabilities.
Solution Approach 2:
The system applies local quality by customizing network optimization parameters according to each device's air interface capabilities. LTE devices receive locally optimized profiles with parameters tailored to their lower throughput capabilities (e.g., smaller buffers, more conservative congestion control) while 5G devices receive profiles optimized for their high-speed capabilities, thereby preventing performance degradation in either group.
Data Source
AI summary
A method, a device, and a non-transitory storage medium are described in which a network performance optimization service is provided. A load balancer device may receive, from an end device, an application service request for an application service that is available from multiple server devices of an application layer network. The load balancer device may determine, from a source identifier associated with the end device and obtained from the application service request, that the source identifier does not map to a network traffic forwarding rule of a set of stored network traffic forwarding rules. In response, the load balancer device may map the source identifier to an application service profile of a set of stored application service profiles; select, based on the application service profile, a first server device of the multiple server devices and establish the first server device to be the destination of the application service request.