Adaptive Cloud Resource Allocation for vRAN Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloud-computing environments face challenges in efficiently managing resources to meet performance goals for workloads, particularly in mobile edge computing, due to issues like insufficient resources, resource interference, hardware accelerator waste, and incorrect cell boundaries, which can be exacerbated by changes in the radio environment.
Innovation Solution
A computer-implemented method for reallocating resources among nodes in a cloud-computing environment by monitoring performance metrics, transferring non-vRAN workloads, adjusting hardware accelerator resources, and adjusting transmit power of vRAN workloads to ensure performance goals are met, utilizing machine learning for decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resources are statically allocated to vRAN workloads, then performance goals can be met, but resource utilization efficiency deteriorates due to insufficient resources during low traffic and waste during high traffic
Solution Approach 1:
The patent implements dynamic resource allocation where the resource allocation manager continuously monitors performance metrics and traffic patterns, then adjusts the allocation of compute resources, storage resources, and network resources to vRAN workloads in real-time. This dynamic approach allows resources to be scaled up during high traffic periods and scaled down during low traffic periods, simultaneously ensuring performance goals are met while optimizing resource utilization efficiency.
2Productivity
If hardware accelerators are dedicated to specific workloads, then performance is improved, but resource waste increases when workloads do not require full accelerator capacity
Solution Approach 1:
The patent implements a shared hardware accelerator pool where multiple vRAN workloads can access the same physical accelerators through virtualization. The resource allocation manager dynamically assigns accelerator resources to different workloads based on real-time performance requirements and traffic patterns. This multi-functional approach allows a single hardware accelerator to serve multiple workloads sequentially, improving performance when needed while eliminating waste during low-utilization periods.
3Extent of automation
If manual resource management is used, then control over resource allocation is maintained, but automation level and response time to changing conditions deteriorate
Solution Approach 1:
The patent implements a closed-loop feedback system where the resource allocation manager continuously monitors performance metrics from vRAN workloads and network conditions, compares actual performance against target performance goals, and automatically adjusts resource allocation accordingly. This feedback mechanism enables high-level automation that responds dynamically to changing conditions while maintaining control through policy-based management, balancing automation benefits with manageable system complexity.
Data Source
AI summary
Described are examples for monitoring performance metrics of one or more workloads in a cloud-computing environment and reallocating compute resources based on the monitoring. Reallocating compute resources can include migrating workloads among nodes or other resources in the cloud-computing environment, reallocating hardware accelerator resources, adjusting transmit power for virtual radio access network (vRAN) workloads, and/or the like.