slides13 - More scheduling

Question 1

Multi-level feedback queue scheduling

Answer 1

• Queues:
  1. real-time processes
  2. system processes
  3. interactive processes
  4. batch processes
• Jobs that use CPU less have higher priority
• use RR for higher queues/FCFS for low queues
• Next queue only runs when previous queue is empty
• Upgrade if process takes less time than quantum/Downgrade if it takes more

Question 2

Real-time processes

Answer 2

1. Soft real-time scheduling: maximize deadline compliance
2. Hard real-time scheduling: deadlines cannot be missed
   - Each process has a period (they use CPU every period), a deadline that must be respected and a burst time (time they use on CPU)

Question 3

Rate monotonic scheduling

Answer 3

• Max CPU utilization: N(2^(1/N) -1) where N is the number of processes (eg. if N = 1-> 100% CPU utilization)
• Preemptive: an incoming process with higher priority will kick out low priority process
• Processes with lower periods assigned higher priority
• CPU utilization: SUM of each processes (burst/period)
• Period = deadline

Question 4

Earliest deadline first scheduling

Answer 4

• Earlier deadline = higher priority
• Preemption when a task with an earlier deadline arrives

Question 5

Proportional share scheduling

Answer 5

• Allocate T shares among all applications/users. If an application/user gets N shares out of T, it gets N/T share of processor time.