Lecture05.pdf - Distributed Process Scheduling Feng Chen...

1
Feng Chen
[email protected]
* Adapted from OSC slides by Silberschatz, Galvin and Gagne, DOSA by Chow and Johnson, and OS slides by B. B Karki, LSU
Distributed Process Scheduling
1.2
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Scheduling Basics
Before execution/running, processes must be scheduled for CPUs and
allocated with resources
Objectives of scheduling
Enhance overall system performance metrics such as process
completion (turnaround) time and processor (resource) utilization
Achieve location and performance transparency
Several challenges:
Multiple processing nodes are geographically distributed

Dynamical behaviors of the system – underlying architecture
Local and global scheduling
Distributed processes execute on remote nodes and may migrate
from node to node:

Remote execution, process/task migration, data migration
Non-negligible communication overhead
1
2

2
1.3
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Process/CPU Scheduling
Scheduler selects from among the processes in ready queue, and
allocates the available processor/CPU to one of them
Scheduling decision may take place
When a process switches state (e.g., running to waiting)
When a new process enters the system
Whenever a scheduling condition is met
When a process terminates
Preemptive vs. non-preemptive
Scheduling can be non-preemptive (static) – once CPU is
allocated to a process, the process keeps CPU until its execution
finishes
Scheduling can be preemptive – process is interrupted
(suspended) for other process to be allocated to CPU
Dispatch latency – time the dispatcher takes to stop one process and
start another running
1.4
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Representation of Process Scheduling
Queuing diagram represents queues, resources, flows
3
4

3
1.5
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Common Scheduling Algorithms
Several scheduling algorithms exist for single processor computer system
First-Come, First Served (FCFS) scheduling
Shortest-Job-First (SJF) scheduling
Shortest-Remaining-Time-First (SRTF) scheduling (preemptive)
Priority scheduling (preemptive/non-preemptive)
Round Robin (RR) scheduling (preemptive)
Multilevel Queue scheduling
Multilevel Feedback Queue scheduling
1.6
Feng Chen, LSU CSC 7103 Advanced Operating Systems
First- Come, First-Served (FCFS) Scheduling
Process Burst Time
P
1
24
P
2
3
P
3
3
Suppose that the processes arrive in the order: P
1
, P
2
, P
3
The Gantt Chart for the schedule is:
Waiting time for P
1
= 0; P
2
= 24; P
3
= 27
Average waiting time: (0 + 24 + 27)/3 = 17
Average turn-around time: (24+27+30)/3=27
P P P
1 2 3
0 24 30 27
5
6

4
1.7
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Optimization Criteria
Speedup – Increasing throughput (the number of completed processes
per time unit)
Resource utilization – keeping processors as busy as possible
Makespan or completion time – decreasing turnaround time (waiting +
execution + event)
Response time – reducing the time taken from request submission until
the first response (important in time-sharing systems)
Load sharing and balancing – key in distributed and multiprocessor
systems
Optimize the max or min values or average measure or variance
1.8
Feng Chen, LSU CSC 7103 Advanced Operating Systems
Multiple Processor Scheduling
Scheduling becomes more complex when multiple CPUs/processors are
available – load sharing/balancing is important
Homogeneous (identical) processors within a multiprocessor system :
Asymmetric multiprocessing

One single processor (called master server like coordinator) does