This avoids stalling the event loop in case of lengthy I/O (network shares, spinning up hard drives etc).