This paper introduces a video application-aware cross-layer framework for joint performance-energy optimization,considering the scenario of multiple users upstreaming real-time Motion JPEG2000 video streams to the acc...This paper introduces a video application-aware cross-layer framework for joint performance-energy optimization,considering the scenario of multiple users upstreaming real-time Motion JPEG2000 video streams to the access point of a WiFi wireless local area network and extends the PHY-MAC run-time cross-layer scheduling strategy that we introduced in (Mangharam et al., 2005; Pollin et al., 2005) to also consider congested network situations where video packets have to be dropped. We show that an optimal solution at PHY-MAC level can be highly suboptimal at application level, and then show that making the cross-layer framework application-aware through a prioritized dropping policy capitalizing on the inherent scalability of Motion JPEG2000 video streams leads to drastic average video quality improvements and inter-user quality variation reductions of as much as 10 dB PSNR, without affecting the overall energy consumption requirements.展开更多
We design a task mapper TPCM for assigning tasks to virtual machines, and an application-aware virtual machine scheduler TPCS oriented for parallel computing to achieve a high performance in virtual computing systems....We design a task mapper TPCM for assigning tasks to virtual machines, and an application-aware virtual machine scheduler TPCS oriented for parallel computing to achieve a high performance in virtual computing systems. To solve the problem of mapping tasks to virtual machines, a virtual machine mapping algorithm (VMMA) in TPCM is presented to achieve load balance in a cluster. Based on such mapping results, TPCS is constructed including three components: a middleware supporting an application-driven scheduling, a device driver in the vip OS kernel, and a virtual machine scheduling algorithm. These components are implemented in the user space, vip OS, and the CPU virtualization subsystem of the Xen hypervisor, respectively. In TPCS, the progress statuses of tasks are transmitted to the underlying kernel from the user space, thus enabling virtual machine scheduling policy to schedule based on the progress of tasks. This policy aims to exchange completion time of tasks for resource utilization. Experimental results show that TPCM can mine the parallelism among tasks to implement the mapping from tasks to virtual machines based on the relations among subtasks. The TPCS scheduler can complete the tasks in a shorter time than can Credit and other schedulers, because it uses task progress to ensure that the tasks in virtual machines complete simultaneously, thereby reducing the time spent in pending, synchronization, communication, and switching. Therefore, parallel tasks can collaborate with each other to achieve higher resource utilization and lower overheads. We conclude that the TPCS scheduler can overcome the shortcomings of present algorithms in perceiving the progress of tasks, making it better than schedulers currently used in parallel computing.展开更多
As data are growing rapidly in data centers,inline cluster deduplication technique has been widely used to improve storage efficiency and data reliability.However,there are some challenges faced by the cluster dedupli...As data are growing rapidly in data centers,inline cluster deduplication technique has been widely used to improve storage efficiency and data reliability.However,there are some challenges faced by the cluster deduplication system:the decreasing data deduplication rate with the increasing deduplication server nodes,high communication overhead for data routing,and load balance to improve the throughput of the system.In this paper,we propose a well-performed cluster deduplication system called AR-Dedupe.The experimental results of two real datasets demonstrate that AR-Dedupe can achieve a high data deduplication rate with a low communication overhead and keep the system load balancing well at the same time through a new data routing algorithm.In addition,we utilize application-aware mechanism to speed up the index of handprints in the routing server which has a 30%performance improvement.展开更多
文摘This paper introduces a video application-aware cross-layer framework for joint performance-energy optimization,considering the scenario of multiple users upstreaming real-time Motion JPEG2000 video streams to the access point of a WiFi wireless local area network and extends the PHY-MAC run-time cross-layer scheduling strategy that we introduced in (Mangharam et al., 2005; Pollin et al., 2005) to also consider congested network situations where video packets have to be dropped. We show that an optimal solution at PHY-MAC level can be highly suboptimal at application level, and then show that making the cross-layer framework application-aware through a prioritized dropping policy capitalizing on the inherent scalability of Motion JPEG2000 video streams leads to drastic average video quality improvements and inter-user quality variation reductions of as much as 10 dB PSNR, without affecting the overall energy consumption requirements.
基金Project (No. 2007AA010305) supported by the National High-Tech R&D Program (863) of China
文摘We design a task mapper TPCM for assigning tasks to virtual machines, and an application-aware virtual machine scheduler TPCS oriented for parallel computing to achieve a high performance in virtual computing systems. To solve the problem of mapping tasks to virtual machines, a virtual machine mapping algorithm (VMMA) in TPCM is presented to achieve load balance in a cluster. Based on such mapping results, TPCS is constructed including three components: a middleware supporting an application-driven scheduling, a device driver in the vip OS kernel, and a virtual machine scheduling algorithm. These components are implemented in the user space, vip OS, and the CPU virtualization subsystem of the Xen hypervisor, respectively. In TPCS, the progress statuses of tasks are transmitted to the underlying kernel from the user space, thus enabling virtual machine scheduling policy to schedule based on the progress of tasks. This policy aims to exchange completion time of tasks for resource utilization. Experimental results show that TPCM can mine the parallelism among tasks to implement the mapping from tasks to virtual machines based on the relations among subtasks. The TPCS scheduler can complete the tasks in a shorter time than can Credit and other schedulers, because it uses task progress to ensure that the tasks in virtual machines complete simultaneously, thereby reducing the time spent in pending, synchronization, communication, and switching. Therefore, parallel tasks can collaborate with each other to achieve higher resource utilization and lower overheads. We conclude that the TPCS scheduler can overcome the shortcomings of present algorithms in perceiving the progress of tasks, making it better than schedulers currently used in parallel computing.
基金the National High Technology Research and Development Program(863)of China(No.2013AA013201)the National Natural Science Foundation of China(Nos.61025009,61232003,61170288 and 61332003)
文摘As data are growing rapidly in data centers,inline cluster deduplication technique has been widely used to improve storage efficiency and data reliability.However,there are some challenges faced by the cluster deduplication system:the decreasing data deduplication rate with the increasing deduplication server nodes,high communication overhead for data routing,and load balance to improve the throughput of the system.In this paper,we propose a well-performed cluster deduplication system called AR-Dedupe.The experimental results of two real datasets demonstrate that AR-Dedupe can achieve a high data deduplication rate with a low communication overhead and keep the system load balancing well at the same time through a new data routing algorithm.In addition,we utilize application-aware mechanism to speed up the index of handprints in the routing server which has a 30%performance improvement.
