Optimizing Soft Real-Time Scheduling Performance for Virtual Machines with SRT-Xen

Abstract

Multimedia applications are an important part of today's Internet. However, currently most virtualization solutions, including Xen, lack adequate support for soft real-time tasks. Soft real-time applications, e.g. media workloads, are impeded by components of virtualization, such as the increase of scheduling latency. This paper focuses on improving scheduling scheme to support soft real-time workloads in virtualization systems. In this paper, we present an enhanced scheduler SRT-Xen. SRT-Xen can promote the soft real-time domain's performance compared with Xen's existing scheduling. It focuses on not only bringing a new realtime-friendly scheduling framework with corresponding strategies but also improving the management of the virtual CPUs' queueing in order to implement a fair scheduling mechanism for both real-time and non-real-time tasks. Finally, we use PESQ (Perceptual Evaluation of Speech Quality) and other benchmarks to evaluate and compare SRT-Xen with some other works. The results show that SRT-Xen supports soft real-time domains well without penalizing non-real-time ones. 2015 15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing 978-1-4799-8006-2/15 $31.00

Lightweight Security Architecture Based on Embedded Virtualization and Trust Mechanisms for IoT Edge Devices

“…3E). Different authors have addressed the hypervisor's scheduler as a way to improve temporal separation and to honor real-time constraints [9]. Additionally, system interrupts require attention since they interfere directly with the VM's execution.…”

Biobjective VoIP Service Management in Cloud Infrastructure

“…Cheng et al (2015) [18] present and compare the SRT-Xen scheduler with other four schedulers (Credit, Credit2, rtglobal, and rtpartition). They focus on real-time-friendly scheduling to improve the management of the virtual CPUs' queueing.…”

“…We also introduce the time-aware versions of RR and WFit strategies (RR 05, RR 10, RR 15, WFit 05, and WFit 10), where we do not allocate calls to VMs in which (2) assigned ← false (3) node index ← 1 (4) Do (5) nodevoice ← get(VNlist, node index) (6) Add call to node voice (7) if utilization of node voice <= 0.7 then (8) assigned ← true (9) else (10) remove call from node voice (11) n o d eindex ← node index + 1 (12) endif (13) While (size of VNlist >= node index and (14) assigned=false) (15) If assigned = false then (16) Create new node voice (17) Add call to new node voice (18) Insert new node voice into VNlist rented time is finished in certain threshold. By these thresholds, we try to avoid next hour renting due to continuation of this call over the rented hour.…”

Optimization of Task Scheduling Algorithm through QoS Parameters for Cloud Computing

“…Provisioning of suitable resources to cloud workload relies upon nature of service necessities of applications. Cloud infrastructure should also support other applications such as big data and soft real-time applications which are the base of today's internet [6]. Where all the scheduling algorithms aim at improving the performance of system and maximizing the profit of service provider, customer satisfaction and feed back is also important.…”