SchedulingaLargeDataCenterCliffSteinColumbiaUniversi.pptVIP

Scheduling a DataCenter Companies run large datacenters Construction, maintainence, etc. of datacenters has significant cost, and uses a significant amount of power Managing such a data center efficiently is an important problem An abstraction of a computing environment Users submit jobs consisting of tasks. Tasks are the unit that is scheduled. Mix of long-running and short-running jobs. Mix of user-facing and back-end jobs. Mix of high and low priority jobs. We will consider a “datacenter” with thousands of machines, and a time period (“day”) long enough to have hundreds of thousands of tasks. The goal We want to evaluate the performance of many different scheduling algorithms on a large datacenter and compare performance Goal: improve cells utilization and overall productivity Meta-goal How does one actually carry out such an experiment? Some ways to measure scheduling quality Throughput - number of processed tasks Total flow time – total time tasks spend in system Total useful work – total time tasks spend processing work that will not be thrown away Number of preemptions – times tasks were interrupted. Pending queue size – number of tasks in system but not being scheduled Machine fragmentation – roughly the number of unused machines Primary Goals Increase throughput. Reduce machine fragmentation (increase job packing ability). Increase the number of unused machines (potential for power savings). Overview We collected data from google “datacenters” We built a high-level model of the scheduling system We experimented with various algorithms How to model machines and jobs Machines: Disk Memory CPU Jobs Consist of set of tasks, which have Cpu, disk, memory, precedence, priority, etc. Processing times Long list of other possible constraints Simulator Replay a “day” of scheduling using a different algorithm. Use data gathered from checkpoint files kept by the scheduling system We tried 11 different alg