Engines are certified by authorities after passing certain test parameters.
The CFM56-7B is certified for thrust rating of 27,300 lbs (121kN).
The engine can be used in several 737 variants, i.e. 600 / 700 / 800.
Due to the obvious differences in the MTOW the max certified rating is not required on all variants, thus a reduced thrust version is available for them.
The so called "rated thrust" is achieved by simply programming a plug in setting pins in a specific order; this is a maintenance job. The engine frame plate will have to be changed as well to reflect to the actual thrust rating.
The N1 ref bug will show you, if you programmed your FMC correctly, the N1 required to provide the full rated thrust reduced for pressure altitude and ambient temperature.
An interesting feature which is not commonly known is the fact that on some 737 you may be able to get some extra thrust.
Let's assume you have an 800 with a 24k rated engine. The 800 is certified for 26k (the 27k is just a non-standard bump power which is not available to all operators). In case you fire-wall the throttles to the full forward stop, you will get 26k power out of the 24k rated engine, this is called emergency reserve thrust.
However, you cannot get 26k emergency reserve thrust on a 600....the 737 600 is not certified for more than 22k.
Confusion perfect?
Example, ferry flight with an 800, FMC set for derate to 24k and assumpted 50 degrees, after pressing TOGA you are looking at the thrust you have programmed.
Pressing TOGA twice will give you the full 24k thrust.
Firewalling the thrust levers will give you the full 26k......bypassing the rating plug.
This works also with engines which are rated at 24k on a fuselage which is certified for the higher rating of 26k.
The N1 bugs, however, will still show you the computer values for the 24k....and this in both cases!
And this is one of the reasons why some companies do not allow the de-rating of the thrust for take-off, only the assumed temperature procedure is used. This way you will never be confused with the actual N1 output being considerably higher than the N1 ref bug show as limit.
But I guess they did not tell this reason to anyone....