How can scheduling algorithms improve system responsiveness?

1 Preemptive vs non-preemptive

One of the main distinctions between scheduling algorithms is whether they are preemptive or non-preemptive. A preemptive algorithm can interrupt a running process and switch to another one, while a non-preemptive algorithm can only switch when the current process finishes or voluntarily yields the CPU. Preemptive algorithms can improve system responsiveness by allowing higher-priority or interactive processes to run sooner, but they also incur more overhead and complexity due to context switching and synchronization. Non-preemptive algorithms can reduce the overhead and ensure that each process gets a fair share of the CPU, but they can also cause long waiting times and poor responsiveness for some processes.

2 Round-robin scheduling

Round-robin scheduling is a simple and widely used preemptive algorithm that assigns each process a fixed amount of CPU time, called a quantum, and cycles through them in a circular order. It can achieve good responsiveness and fairness for processes with similar characteristics, but it can also suffer from high context switching overhead and poor performance for processes with different CPU and I/O demands. To improve round-robin scheduling, some variations have been proposed, such as dynamic quantum allocation, priority-based scheduling, and multilevel feedback queues.

4 Rate-monotonic scheduling

Rate-monotonic scheduling (RMS) is a preemptive algorithm that is designed for real-time systems, where processes have strict deadlines and periodic execution. It assigns the highest priority to the process with the shortest period, and preempts any lower-priority process if a higher-priority one becomes ready. It can guarantee that all processes meet their deadlines, if the system is schedulable, meaning that the total CPU utilization does not exceed a certain limit. However, it can also cause priority inversion, where a higher-priority process is blocked by a lower-priority one that holds a shared resource. To prevent this, priority inheritance or priority ceiling protocols can be used.

5 Lottery scheduling

Lottery scheduling is a probabilistic algorithm that assigns each process a number of tickets, which represent its share of the CPU. It randomly draws a ticket to select the next process to run, and repeats this process every quantum. It can achieve proportional fairness and responsiveness, as well as adapt to changing process characteristics and system load. However, it can also introduce randomness and unpredictability, as well as require extra mechanisms to handle ticket transfer, compensation, and inflation.