This question really is not useful as-is, not to the community, and not even to yourself.
You are essentially asking us to choose a load balancing and/or high availability solution for your service, without providing any information about the service, and quite possibly without even having built a prototype of the service!
This is not going to work.
Every service has its own unique characteristics which make certain LB/HA strategies more useful, or even necessary, and so without knowing much more about the design of the service it's really impossible to give a good answer.
As it stands, any attempts at answers would either make a recommendation likely to be bad, or break the post length limit trying to explain basic concepts, or possibly both.
Before you move forward, make sure that you have a working service with five users: your group of developers at your startup, or wherever you are, should be [dogfooding][1] it. Only when you have both a minimum viable product and a growing userbase should you even begin to think about scaling it.
And if you haven't done so already, you should study existing open source solutions very carefully. You probably should be using one. [1]: https://en.wikipedia.org/wiki/Eating_your_own_dog_food