Performance Prediction Based Workload Scheduling in Co-Located Cluster

Cloud service providers generally co-locate online services and batch jobs onto the same computer cluster,where the resources can be pooled in order to maximize data center resource utilization.Due to resource competition between batch jobs and online services,co-location frequently impairs the performance of online services.This study presents a quality of service(QoS)prediction-based schedulingmodel(QPSM)for co-locatedworkloads.The performance prediction of QPSM consists of two parts:the prediction of an online service’s QoS anomaly based on XGBoost and the prediction of the completion time of an offline batch job based on randomforest.On-line service QoS anomaly prediction is used to evaluate the influence of batch jobmix on on-line service performance,and batch job completion time prediction is utilized to reduce the total waiting time of batch jobs.When the same number of batch jobs are scheduled in experiments using typical test sets such as CloudSuite,the scheduling time required by QPSM is reduced by about 6 h on average compared with the first-come,first-served strategy and by about 11 h compared with the random scheduling strategy.Compared with the non-co-located situation,QPSM can improve CPU resource utilization by 12.15% and memory resource utilization by 5.7% on average.Experiments show that the QPSM scheduling strategy proposed in this study can effectively guarantee the quality of online services and further improve cluster resource utilization.

Quantitative analysis of real-time performance and hardware requirements for edge computing platform

For real-time edge systems such as autonomous driving,not only the correctness of task functions,but also the response and processing time of tasks should be satisfied.In the hardware selection phase of a real-time system,time series analyses must be performed on the hardware platform running real-time applications.At present,the common method of worst-case execution time(WCET)analysis focuses mainly on analyzing the impact of hardware platform architecture or task execution process on the task running time.However,different tasks in an autopilot system have different levels of urgency,and preemption between tasks is the main factor that affects the task execution time.The key problem is how to quantify the time fluctuation caused by task preemption for each subtask of the autopilot system running on a fixed hardware platform.This paper presents a time analysis method for a real-time application based on a queuing theory and preemptive scheduling strategy,which assigns different priorities to tasks according to their time urgency and preemptive scheduling according to task priority.Through an experimental case study,the impact of the running time of each subtask in a real-time application with task priority preemptive scheduling is analyzed,along with the impact of changes in hardware platform performance on such real-time applications.