The end desired result is a more performant system with less resource utilisation. I want to connect a sql database to cloud functions and host a set pf api's.

Here are two alternatives...

Alternative one: We host each api as a separate cloud function.

const functions = require('firebase-functions');

exports.helloFunc1 = functions.https.onRequest(function (request, response){
 response.send("Hello from Func1!");
});

exports.helloFunc2 = functions.https.onRequest(function (request, response){
 response.send("Hello from Func2!");
});

Alternative two: We host a single cloud function with internal routing.

const functions = require('firebase-functions');
const express = require('express');
const router = new express.Router();

var app = express();

var helloFunc1 = function (request, response){
 // After some DB OPS
 response.send("Hello from Func1!");
};

var helloFunc2 = function (request, response){
 // After some DB OPS
 response.send("Hello from Func2!");
};


router.get('/helloFunc1',helloFunc1);
router.get('/helloFunc2',helloFunc2);

exports.root = functions.https.onRequest(router);

Help me understand the tradeoffs between the approaches and any advantages/disadvantages in either designs.

Also please consider the database connection pooling as a part of the analysis. This wont be a determining factor if we use firebase or datastore, but for native SQL Databases, i'm assuming a lot of connections will impact performance when this is auto scaled.

EDIT #1

This is a rudimentary example, but in reality the implications maybe a lot more. Is working with a framework which provides additional functionality like ORM's and shared models, Middleware for AUTH, general convention over configuration approach a real option in a server-less environment? Because we know that these frameworks are designed to run on a always-up server. How will ORM's & frameworks in general deal with being preempted and how will it affect performance?

PS. I'm new to Node and Express.