Results 1 -
3 of
3
D-factor: a quantitative model of application slow-down in multi-resource shared systems
- in Proceedings of the 12th ACM SIGMETRICS/PERFORMANCE conference
, 2012
"... ABSTRACT Scheduling multiple jobs onto a platform enhances system utilization by sharing resources. The benefits from higher resource utilization include reduced cost to construct, operate, and maintain a system, which often include energy consumption. Maximizing these benefits comes at a price -re ..."
Abstract
-
Cited by 7 (0 self)
- Add to MetaCart
(Show Context)
ABSTRACT Scheduling multiple jobs onto a platform enhances system utilization by sharing resources. The benefits from higher resource utilization include reduced cost to construct, operate, and maintain a system, which often include energy consumption. Maximizing these benefits comes at a price -resource contention among jobs increases job completion time. In this paper, we analyze slow-downs of jobs due to contention for multiple resources in a system; referred to as dilation factor. We observe that multiple-resource contention creates non-linear dilation factors of jobs. From this observation, we establish a general quantitative model for dilation factors of jobs in multi-resource systems. A job is characterized by a vector-valued loading statistics and dilation factors of a job set are given by a quadratic function of their loading vectors. We demonstrate how to systematically characterize a job, maintain the data structure to calculate the dilation factor (loading matrix), and calculate the dilation factor of each job. We validate the accuracy of the model with multiple processes running on a native Linux server, virtualized servers, and with multiple MapReduce workloads co-scheduled in a cluster. Evaluation with measured data shows that the D-factor model has an error margin of less than 16%. We also show that the model can be integrated with an existing on-line scheduler to minimize the makespan of workloads.
Migration, assignment, and scheduling of jobs in virtualized environment
- In HotCloud
, 2010
"... Abstract Migration is an interesting issue for managing resource utilization and performance in clusters. Recent advances in server virtualization have made migration a practical method to achieve these goals. Especially, the live migration of virtualized servers made their pausing times negligible ..."
Abstract
-
Cited by 5 (1 self)
- Add to MetaCart
(Show Context)
Abstract Migration is an interesting issue for managing resource utilization and performance in clusters. Recent advances in server virtualization have made migration a practical method to achieve these goals. Especially, the live migration of virtualized servers made their pausing times negligible. However, migration of a virtual machine (VM) can slow down other collocated VMs in multiresource shared systems, where all the system resources are shared among collocated VMs. In parallel execution environment, such sudden slow-down phase of systems is called system noise; it may slow down overall systems while increasing the variability of system performance. When we consider the virtual machine assignment problem as resource allocation, those performance issues are hard to be properly treated. In this work, we address how to consider performance in assigning VMs. To achieve this goal, we model a migration process of a VM instance as a pair of jobs that run at the hosts of sender and receiver. We propose a method to analyze the migration time and the performance impact on multiresource shared systems for completing given VM assignment plan. This study may contribute to create more robust performance in virtualized environment.
Insoon Jo et al. / International Journal on Computer Science and Engineering (IJCSE) Workload-aware VM Scheduling on Multicore Systems
"... Abstract—In virtualized environments, performance interference between virtual machines (VMs) is a key challenge. In order to mitigate resource contention, an efficient VM scheduling is positively necessary. In this paper, we propose a workload-aware VM scheduler on multi-core systems, which finds a ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract—In virtualized environments, performance interference between virtual machines (VMs) is a key challenge. In order to mitigate resource contention, an efficient VM scheduling is positively necessary. In this paper, we propose a workload-aware VM scheduler on multi-core systems, which finds a systemwide mapping of VMs to physical cores. Our work aims not only at minimizing the number of used hosts, but at maximizing the system throughput. To achieve the first goal, our scheduler dynamically adjusts a set of used hosts. To achieve the second goal, it maps each VM on a physical core where the physical core and its host most sufficiently meet the resource requirements of the VM. Evaluation demonstrates that our scheduling ensures efficient use of data center resources. Keywords- server consolidation; virtualization; virtual machine scheduling; multi-core systems.
