Presentation is loading. Please wait.

Presentation is loading. Please wait.

Computing Resource Allocation and Scheduling in A Data Center

Similar presentations


Presentation on theme: "Computing Resource Allocation and Scheduling in A Data Center"— Presentation transcript:

1 Computing Resource Allocation and Scheduling in A Data Center
2016/03/30

2 Background Tasks from users are processed by server(s) in a data center. Wrapped as virtual machines or containers. Resource allocator in a data center deploys tasks to servers upon their arrival. Scheduler in a server determines which task to be executed next.

3 Motivation The current designs of resource allocator do not cooperate with the scheduler. Vise versa. The heterogeneity of hardware and varying behaviors of applications increase the difficulty of generating good resource allocations and schedules. With guaranteed QoS, energy-efficiency, … etc.

4 Goal Design a hierarchical resource management system.
Integrate resource allocator and scheduler. Improve the resource utilization and energy-efficiency of servers in a data center.

5 Current Project Resource allocation Scheduling
Manage the number of containers and their deployments for each application in a data center. Work with CHT. Scheduling Energy-efficient task scheduler on asymmetric multi-core platform.

6 Resource Allocation Based on Linux container technique.
Docker / Kubernetes. Design and implement two components: Container number manager adjusts the number of containers according to application performance. Resource allocator determines which server a newly created container should run on.

7 Scheduling A scheduler that determines the frequency of cores and job-to-core assignment in order to reduce energy consumption. Currently target on throughput guaranteed jobs.

8 Future Directions A hierarchical resource management system that integrates the resource allocator and the scheduler. Design a scheduler that takes task characteristics into consideration. Information of task characteristics comes from the resource allocator. Design a resource allocator that takes task runtime information into consideration. Runtime information comes from the scheduler.

9 Future Directions(Cont.)
An interface for exchanging the information between resource allocator and scheduler(s).

10 Discussion


Download ppt "Computing Resource Allocation and Scheduling in A Data Center"

Similar presentations


Ads by Google