What are the best practices to improve JuMP perfomance?


I’ve been working in a huge LP problem. My problem can easily overcome 1million of variables and constraints. As I was not familiar to Julia I have coded in a global scope called “Main.jl” and it worked well. My mainly objects are data frames and dictionaries. Both help me to build my optimization model. I have no function, just a code with around 3k lines. I have read the documentation about performance in Julia and I have noticed that it’s better to build functions in different files and call them from another file (i.e, a “main”). I also could create a module and organize my functions there. However, I also read some comments that said the JuMP speed can not be so good when we use many include() statements. I’d like to organize my code in functions in order it can match the best practices of development. Anyone could help me with any advice?Thanks in advance.