If you're going to write your own, you may want to check into uCOS from Jean Labrosse. I think someone has already ported it to the 8051.

If you want to BUY a solution, you can get one from Keil Software. We have a tiny real-time operating system (RTX51 Tiny) for the 8051. It supports round-robin and cooperative multitasking. Tasks (and interrupts) may send a signal to another task. Tasks may also delay for a number of clock ticks or for a specified interval. RTX51 Tiny takes up about 800 bytes of code space and requires NO XDATA. You can get more information about it at

http://www.keil.com/rtx51

BTW. Full assembly source code is included.

Hope that helps,
Jon Ward
Keil Software