
A user-level scheduler that replaces the heavy model of system threads and I/O functions with lighter fibers and more efficient I/O models without changing the application code.
This can significantly reduce the overhead of the operating system. We will discuss how such a substitution is implemented in practice and what performance gains it provides. At the center of this talk is a universal approach to accelerating multithreaded and asynchronous systems without magic, but with a visible effect.
Speakers


Dmitry Olshansky
Huawei
