Large-scale software systems like Google - deployed over a world-wide network of hundreds of thousands of computers - have become a part of our lives. These are systems success stories - they are reliable, available ("up" nearly all the time), handle an unbelievable amount of load from users around the world, yet provide virtually instantaneous results. On the other hand, many computer systems don't perform nearly as well as Google - hence the now-cliche "the system is down." In this class, we study the scientific principles behind the construction of high-performance, scalable systems. The course begins with a discussion of the relevant features of modern architectures, and moves up the stack from there to operating system services such as programming language runtime systems, concurrency and synchronization, with a focus on key operating system features, I/O and networking, and distributed services.
Monday, March 15, 2021 to Sunday, June 20, 2021