In the interest of promoting real learning, I'll offer some suggestions without giving too much away.
'Taking 20K off the top of a file' isn't terribly useful as an instruction. What's in the file? There are few formats that are structured around that kind of boundary. Try this trick with a jpeg and let me know what it looks like
'Take 20K
lines off the top of a file' make more sense because you are dealing with probable record boundaries.
Here are a few more tools to do this stuff (not an exaustive list):
head
tail
split
cut
and don't forget awk, sed and perl.
man those and have fun!