Supporting soft real-time tasks in the xen hypervisor

Lee, Min; Krishnakumar, A. S.; Krishnan, P.; Singh, Navjot; Yajnik, S.

doi:10.1145/1735997.1736012

Cited by 112 publications

(49 citation statements)

References 23 publications

(19 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Cheng et al [14] provide a dual run-queue scheduler: for VCPUs with periodic outgoing packets, they are scheduled in a Earliest Deadline First queue, while for other VCPUs they are still scheduled in the default credit queue. Lee et al [12] patched the Credit scheduler so it will boost not only blocked VCPUs, but also active VCPUs (so that if a VCPU also runs a background CPU intensive task, it can benefit from the boost as well). However, none of these approaches strictly prioritizes VCPUs.…”

Section: Related Workmentioning

confidence: 99%

Prioritizing local inter-domain communication in Xen

et al. 2013

2013 IEEE/ACM 21st International Symposium on Quality of Service (IWQoS)

View full text Add to dashboard Cite

Abstract-As computer hardware becomes increasingly powerful, there is an ongoing trend towards integrating QoS-critical systems as virtual machines (domains) on a common, virtualized computing platform. Given the lower latency of local interdomain communication (IDC) on the same host (compared to inter-host communication), system administrators may preferably colocate domains so that they can communicate locally. When multiple IDC flows contend on the same host, it is important to properly prioritize IDC flows among domains to meet their respective QoS requirements. This paper examines the limitations of IDC in Xen, a widely used open-source virtual machine monitor (VMM) that recently has been extended to support real-time domain scheduling. We find that both the VMM scheduler and the manager domain can significantly impact IDC QoS under different conditions, and show that improving the VMM scheduler alone cannot effectively prevent priority inversion for local IDC. To address those limitations, we present RTCA, a Real-Time Communication Architecture within the manager domain in Xen, along with experimental results that demonstrate the latency of high-priority IDC can be improved dramatically from ms to µs by a combination of the RTCA and a real-time VMM scheduler.

show abstract

Section: Related Workmentioning

confidence: 99%

Prioritizing local inter-domain communication in Xen

et al. 2013

2013 IEEE/ACM 21st International Symposium on Quality of Service (IWQoS)

View full text Add to dashboard Cite

show abstract

“…From experiments in [4], [17], these losses can be further reduced by tweaking scheduler parameters such as weights for Dom0 since Dom0 is usually the bottleneck for I/O intensive workloads.…”

Section: Figure 5 Media Stream Plots In No-net Configurationmentioning

confidence: 99%

Towards high-availability for IP telephony using virtual machines

Patnaik

Bijlani

Singh

2010

2010 IEEE 4th International Conference on Internet Multimedia Services Architecture and Application

View full text Add to dashboard Cite

Abstract-Virtualization technology is popular today for hosting Internet and cloud-based compute services. Recently, the IP Communications domain has seen the adoption of virtualization with enterprise telephony solutions being hosted in virtualized environments. A deployment scenario that is common in the IP communications domain is the virtual appliance. A virtual appliance encapsulates the IP Telephony application, such as a VoIP Call/Proxy Server, and the OS software, within a virtual machine. This allows for a single physical machine to host a number of virtual appliances -each being a telephony application deployed on an OS configuration for which it has been optimized. Apart from the well known benefits of virtualization such as server consolidation and improved resource utilization, a promising capability of virtualization technology is that of live virtual machine migration. In case of failure, a live virtual appliance can be efficiently and transparently migrated to a different physical machine with minimal impact on clients using the hosted IP communication service, thereby providing high-availability and fault tolerance. This work studies the feasibility of deploying IP communications infrastructure on virtualized platforms, with high-availability and reliability as a goal. We model a real IP telephony workload to understand the performance implications of such a deployment, and the effectiveness of these high-availability mechanisms. We use quantitative methods to measure performance such as call throughput, delay and packet loss. By understanding the characteristics of IP telephony workloads, we identify specific aspects that affect performance and also identify some optimal configurations. The results presented in this work will be useful for telecommunication service providers to understand the benefits and limitations of such a deployment.

show abstract

“…The software layer providing the virtualization, inserted between the hardware and VMs, is called the hypervisor or the Virtual Machine Monitor (VMM) [1]. The VMM enables concurrent execution of multiple VMs, isolate them, and schedule them among the available resources.…”

Section: Introductionmentioning

confidence: 99%

“…Also, it has been used for soft real-time applications such as media-based ones which demonstrate inadequate performance due to the impede of virtualization components such as low performance virtualization I/O and increased scheduling latency. [1] Therefore, the performance overhead introduced by virtualization should be limited.…”

Section: Introductionmentioning

confidence: 99%