vRAN Compute Resource Sharing Through Real-Time Workload Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional scheduling frameworks for virtualized Radio Access Networks (vRAN) in telecommunications networks are inefficient and expensive due to overprovisioning of resources during peak capacity, leading to significant resource wastage during non-peak periods, and lack a framework to intelligently share unused CPU cores among different workloads.
Innovation Solution
A real-time scheduling system that utilizes telemetry data to estimate runtime durations for vRAN workloads and generates scheduling instructions to optimize the allocation of computing resources across vCPUs, ensuring deadlines are met while sharing resources with other workloads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If isolated compute resources are overprovisioned to meet peak capacity demands, then service reliability is improved, but resource utilization deteriorates
Solution Approach 1:
The system dynamically adjusts compute resource allocation based on real-time workload conditions. During peak periods, vRAN workloads receive dedicated isolated resources to ensure service reliability. During non-peak periods, the system automatically shares idle compute resources with other workloads, transforming the static overprovisioning model into a dynamic adaptive one that resolves the contradiction between reliability and utilization
Solution Approach 2:
The compute resources are segmented into time-based allocation periods. The system divides resource allocation into peak period segments (where isolated resources guarantee reliability) and non-peak period segments (where shared resources maximize utilization). This temporal segmentation allows the system to satisfy reliability requirements during critical periods while improving overall resource productivity during idle periods
2Reliability
If dedicated compute resources are allocated to vRAN workloads, then processing deadlines are met, but power consumption increases
Solution Approach 1:
The system implements periodic evaluation of workload conditions and adjusts resource allocation accordingly. During periods when vRAN workloads require full resources, dedicated allocation ensures deadline compliance. During periods of lower demand, the system periodically transitions to shared resource mode, reducing power consumption while maintaining the ability to meet deadlines when needed
Solution Approach 2:
The system changes the allocation parameter of compute resources from fixed dedicated allocation to variable allocation based on workload intensity. By monitoring workload parameters and adjusting resource allocation dynamically, the system maintains deadline meeting capability when workload requires it while reducing power consumption during low-utilization periods through shared resource allocation
3Productivity
If compute resources are shared among multiple workloads, then resource utilization improves, but scheduling complexity increases
Solution Approach 1:
The scheduling system employs self-service mechanisms where workload agents autonomously report their resource requirements and readiness states. The scheduler receives structured telemetry data from workloads and automatically makes allocation decisions based on predefined policies and real-time conditions. This self-service approach enables multi-workload resource sharing while keeping scheduling complexity manageable through automation and standardized interfaces
Data Source
AI summary
The present disclosure relates to systems, methods, and computer-readable media for implementing a scheduler for vRAN compute sharing. The systems described herein involve a real-time scheduling system that considers telemetry data associated with usage of vCPUs on a VM, container, or other service construct and determines estimated runtimes for tasks of workloads running on the vCPUs. The real-time scheduling system may generate scheduling instructions to be used by an operating system on the server device to schedule allocation of computing resources to any number of vCPUs hosted by the server device. The real-time scheduling system provides features that enables optimization of not only physical layer processing tasks, but a wholistic approach that involves optimizing scheduling of tasks associated with multiple processing layers of VMs, and particular vRAN workload VMs.