Moreover so as to parallelize the transfer even when blocking I/O is used, the code tries to write down a small quantity of data to every fd in a loop, in order that the kernel will ship the packets in the background to multiple slaves at the. https://xn--L3ca4BWE5B2b.com/