可以去看tornado的ioloop模块。里面有说明

...
class IOLoop(Configurable):
    """A level-triggered I/O loop.
    We use `epoll` (Linux) or `kqueue` (BSD and Mac OS X) if they
    are available, or else we fall back on select(). If you are
    implementing a system that needs to handle thousands of
    simultaneous connections, you should use a system that supports
    either `epoll` or `kqueue`.
...

在Linux系统中用epoll,BSD或Mac OS X用kqueue,其他系统中用select。

使用的是python标准库中的select模块。实际上select模块也只是对系统的select的调用，并没有自己实现。

想深入研究的话可以看源码

感谢 @依云 的补充。
python标准库中的select模块是对系统各种I/O复用方案的封装。

>>> import platform
>>> platform.linux_distribution()
('Red Hat Enterprise Linux Server', '6.5', 'Santiago')
>>> import select
>>> dir(select)
['EPOLLERR', 'EPOLLET', 'EPOLLHUP', 'EPOLLIN', 'EPOLLMSG', 'EPOLLONESHOT', 'EPOLLOUT',         
'EPOLLPRI', 'EPOLLRDBAND', 'EPOLLRDNORM', 'EPOLLWRBAND', 'EPOLLWRNORM', 'PIPE_BUF', 
'POLLERR', 'POLLHUP', 'POLLIN', 'POLLMSG', 'POLLNVAL', 'POLLOUT', 'POLLPRI', 'POLLRDBAND', 
'POLLRDNORM', 'POLLWRBAND', 'POLLWRNORM', '__doc__', '__file__', '__name__', 
'__package__', 'epoll', 'error', 'poll', 'select']