GWU OS: Linux and Real-time Scheduling
In this talk, we focus on scheduling in a couple of systems.
1. Scheduling in a general-purpose OS (Linux) and how it tries to minimize response time, while limiting system overheads. This will show how systems that know very little about their applications attempt to balance the system's competing goals.
2. We'll also discuss how relying on specialized execution patterns (prevalent in real-time systems), enables a more specialized scheduling environment for which global and strong guarantees can be made about execution. These culminate in a mathematical analysis that bounds response times to what they need to by for physical system control.
Please note: @12:19, I flipped the values. Should be 20/3 and 20/6. See @Maksym Planeta in the comments. Thanks for the correction!
0:00 Introduction
1:22 Linux Goals
2:55 CFS - Completely Fair Scheduler
13:15 CFS Scheduling Example
19:29 CFS Summary
23:14 Real-time Scheduling (RTS) Introduction
25:49 Real-time System Examples
34:03 RTS Goals and Model
43:04 RTS Task Execution Examples
51:48 Schedulability and the Response Time Analysis (RTA)
1:08:03 RTA Examples
1:16:26 Expanding the RTA
1:20:52 Comparing the scheduling models