What are we talking about exactly, what and what about it specifically needs to scale in what way?

Does it need to process more things per time?

Does it need to run on more machines?

Is it time critical / is responsiveness critical?

What are it's bottlenecks? Is it mem-bound? CPU-bound? IO-bound?

What is "it"?

Is "it" connecting to something else, and what does "something" run on?