Cloud Streaming Server GPU Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud streaming services face inefficiencies in resource utilization due to reliance on central processing units (CPU) and suboptimal use of graphics processing units (GPUs) for video decoding, leading to potential bottlenecks and increased costs.
Innovation Solution
A cloud streaming server that dynamically determines the optimal GPU for video decoding based on usage rates, selecting the GPU with the lowest usage rate or the main GPU with the best performance to distribute load and maximize resource efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the cloud streaming service uses a single main GPU for video decoding, then the system structure is simple, but resource utilization efficiency deteriorates and bottlenecks occur during peak loads
Solution Approach 1:
The patent implements dynamic GPU selection based on real-time usage rates. The system continuously monitors usage rates of multiple GPUs and dynamically switches between them depending on current load conditions. This dynamic approach allows the system to optimize resource utilization without being constrained by a fixed single-GPU architecture, resolving the contradiction between structural simplicity and productivity.
Solution Approach 2:
The patent divides the video decoding workload across multiple GPUs instead of relying on a single main GPU. By segmenting the processing load and introducing a backup GPU that can be activated when needed, the system achieves better resource distribution and reduces bottlenecks while maintaining manageable system complexity through modular design.
2Ease of operation
If the browser uses its own decoder for video playback, then the service is easy to implement, but system resource optimization deteriorates
Solution Approach 1:
The patent introduces a cloud streaming server as an intermediary between the browser and the video decoding process. Instead of the browser directly decoding videos using its own resources, the server intercepts video streams and redirects them to optimal GPU decoders. This intermediary approach maintains ease of implementation for the browser while achieving superior system resource optimization through centralized GPU management.
3Reliability
If the system always uses the main GPU for video decoding, then the decoding performance is consistent, but the likelihood of bottlenecks and service delays increases during peak loads
Solution Approach 1:
The patent implements a feedback mechanism that continuously monitors GPU usage rates and adjusts the decoding workload accordingly. When the main GPU approaches capacity limits, the system detects this through usage rate monitoring and automatically redirects video decoding tasks to backup GPUs. This feedback-driven dynamic allocation maintains decoding performance consistency while preventing service delays during peak loads.
Solution Approach 2:
The system dynamically switches between main and backup GPUs based on real-time usage conditions rather than statically assigning all decoding tasks to the main GPU. This dynamic flexibility allows the system to maintain reliable decoding performance while avoiding bottlenecks by distributing load to available resources when the main GPU is saturated.
Data Source
Figure 1~3
Figure 4~5
Figure 6
AI summary
A cloud streaming service system, a cloud streaming service method using an optimal GPU, and an apparatus for the same are disclosed. A cloud streaming service can be provided by determining whether a video is played on a browser, when a playback of the video is sensed, decoding the video by using a video decoding entity determined on the basis of system resource conditions, and performing a streaming procedure corresponding to capturing, encoding and sending the decoded video. When a video is provided through the cloud streaming service, it is possible to effectively use the resources of a cloud streaming system by efficiently distributing a plurality of GPUs equipped in the cloud streaming system.