Well, it's not so different than when you have a big job to do.
You can either try to do it all by yourself, or you can get some people to help you.
If you get more people, you need to divide up the job into smaller tasks so that each person can keep busy.
Sometimes, it will take more work to manage the people than just getting on with the job yourself - it all depends on the job.
Some jobs are easily "parallelizable", some are not.
Say you're building a house. You can't work on the foundation, the walls and the roof at the same time. The foundation needs to go in first, then the walls, then the roof. You have to be very clever to get the right things done at the right time, and you can't keep many people busy at the same time.
If you were building 20 houses, that would be different.
At the beginning, you could lay the foundation of the first house.
Then, you work on the walls of the first house while you're laying the foundation of the second house.
Then, you do the roof of the first house, the walls of the second house, and the foundation of the third house.
After that bit of inefficiency at the beginning, by the third step you see that you're building three houses in parallel, with everyone busy.
That's how a pipeline works. This idea comes up often in computer engineering, whether you're writing programs to run on multi-core processors or doing network-distributed parallel processing (Google, etc.).
Parallel processing is not an easy problem, and you can't just install a Beowulf cluster of Linux machines and expect everything to be optimized and running efficiently.