Web Application

Deployment: Heroku
Copyright © 2025 by
Robert M. Dondero, Ph.D
Princeton University

1
 Objectives
• This lecture will cover:
– How to deploy a web app and database to

Heroku

2
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

3
 Deployment Options
• Question:
– Where/how might you run Penny app so it’s
constantly available to the world…
• Using a production-quality HTTP server?
• Using a production-quality DBMS?

4
 Deployment Options
• Answer 1: Your computer

5
 Question (11deployheroku)
• Why would it be unreasonable to make your app
publicly available on your computer? Choose
all that apply:
– (a) Your computer would need to be turned on
all/most of the time
– (b) Your computer would need to be connected to the
Internet all/most of the time
– (c) Your computer would need to be powerful enough
to handle a potentially large volume of requests
– (d) You would need sys admin skills to install &
configure a production-quality web server & DBMS
• Browse to https://cos333attend.cs.princeton.edu
to answer

[1 point] 6
 Deployment Options
• Answer 2: Princeton CS Dept HTTP
servers
– See
https://csguide.cs.princeton.edu/publishing/w
ebpages
– Provides MySQL databases
– Pros and cons:
• (pro) XXX.cs.princeton.edu address
• (pro) Free
• (con) Requires CS Dept approval
• (con) No access to server logs

7
 Deployment Options
• Answer 3: Princeton OIT servers
– Pros and cons:
• (pro) Free
• (con) Requires OIT approval
• (con) Requires sponsorship of some Princeton
department
• (con) Difficult; requires intense cooperation with
OIT

8
 Deployment Options
• Answer 4: Commercial cloud service on
your own
– Render.com, Heroku, DigitalOcean, ...
– Pros and cons:
• (pro) The service does sys admin for you
• (con) You may need to pay for it!

– The best option for most COS 333 projects

9
 Deployment Options

– But which cloud service?

– Must choose a service for:

• Web application
• Database

10
 Deployment Options
A little COS 333 history:

Historical Phase Application Cloud Service DB Cloud Service

1 Heroku (free forever) Heroku (free forever)

Heroku eliminated its free tier

2 Render (free forever) Render (free for 3 mths)

2 Render (free forever) ElephantSQL (free forever)

Heroku teamed with GitHub: GitHub Student Developer Pack

3 Render (free forever) Render (free for 3 mths)

3 Render (free forever) ElephantSQL (free forever)

3 Heroku with GitHub Student Heroku with GitHub Student

Developer Pack (free for 1 yr) Developer Pack (free for 1 yr)

11
 Deployment Options
A little COS 333 history (cont.):

Historical Phase Application Cloud Service DB Cloud Service

Render limited its free tier DBs to 1 month

ElephantSQL went out of the DB business

4 Render (free forever) Render (free for 1 mth)

4 Render (free forever) Neon (free forever)

4 Heroku with GitHub Student Heroku with GitHub Student

Developer Pack (free for 1 yr) Developer Pack (free for 1 yr)

Heroku extended GitHub Student Developer Pack to 2 yrs

5 Render (free forever) Render (free for 1 mth)

5 Render (free forever) Neon (free forever)

5 Heroku with GitHub Student Heroku with GitHub Student

Developer Pack (free for 2 yrs) Developer Pack (free for 2 yrs)

12
 Deployment Options
Web app cloud services:

Web App Cost

Cloud Service
Render.com Free Free for 750 hours of running time per
month *
Render.com Starter $7/month

Heroku Eco $5 for 1000 dyno hours per month **

Heroku Basic ~$0.01 per hour (max of $7 per month)

* Slow deployment; slow at runtime?

** GitHub Student Developer Pack provides $13/month
for 2 years; cannot renew

13
 Deployment Options
Database cloud services (cont.):
DB Cloud Time Size Concurrent Cost
Service Period Connections

Render.com Free 1 month * 1 GB 97 0

Render.com Unlimited 1 GB 97 $7/month

Starter

Neon Free Plan Unlimited 0.5GB 10000 0

Neon Launch Unlimited 10GB 10000 $19/month

Heroku Mini Unlimited 1 GB 20 $5/month **

Heroku Basic Unlimited 10 GB 20 $10/month

* Can create another

** GitHub Student Developer Pack provides $13/month
for 2 years; cannot renew
14
 Deployment Options
• All of those DB options (can) use
PostgreSQL

15
Deployment Options

Michael
Stonebreaker

16
 Deployment Options
• PostgreSQL assessment (vs. SQLite)
– (con) Setup on local computer (not shown) is
much more complex
– (con) Setup on cloud service is more complex
– (pro) Production-quality
• Distinct process
• Authentication
• Row-level (vs. database-level) locking
• Reasonable for apps deployed to the cloud
• …

17
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

18
 Heroku

Orion Adam James

Henry Wiggins Lindenbaum

19
 Heroku
• Render
– Can create app or DB first
• Heroku
– Must create app first
– Then create DB specifically for the app

20
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

21
 Defining the App
• Deployment
Recommended
Heroku Heroku
deploy

GitHub
push

Local Computer Local Computer

22
 Defining the App
• You must change names as appropriate…

23
 Defining the App
• After performing setup steps in Git and
GitHub Primer document...

• Create a GitHub repo

– Browse to https://github.com
• Click the New button
– For Repository name enter pennyheroku
– Click the Private radio button
– Check the Add a README file check box
– Click the Create Repository button

24
 Defining the App
• Clone the GitHub repo to my local
computer

$ git clone https://github.com/rdondero/pennyheroku.git

– Creates pennyheroku dir on local computer

25
 Defining the App
• Create these files in pennyheroku dir:
– From the PennyFlaskJinja app
• runserver.py (optionally)
• penny.sql
• penny.sqlite (optionally)
• book.py
• header.html, footer.html, index.html,
searchform.html, searchresults.html
• penny.py
– Described later in this lecture
• database.py

26
 Defining the App
• Also create in pennyheroku dir:
– requirements.txt
Flask
psycopg2
SQLAlchemy
python-dotenv
gunicorn

• Tells Heroku what additional modules the app uses

• Create manually, or issue command
python –m pip freeze >
requirements.txt

27
 Aside: Gunicorn

WSGI Python
HTTP
Browser HTTP WSGI
Server
Server App
HTTP HTTP WSGI

Chrome Nginx Gunicorn Flask

Firefox Apache Phusion Django
Passenger

28
 Defining the App
• Also create in pennyheroku dir:
– Procfile

web: gunicorn penny:app

• Tells Heroku the app’s process type
– It’s a web app
• Tells Heroku the command to start the process
– Gunicorn (https://gunicorn.org/)

29
 Defining the App
• Also create in pennyheroku dir:
– runtime.txt
python-3.12.3

• Tells Heroku which version of Python it should use

30
 Defining the App
• Stage the app to the local git repo, commit
the app to the local git repo, and push it to
the GitHub repo

$ cd pennyheroku
$ git add .
$ git commit -m "initial load"
$ git push origin main

31
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

32
 Deploying the App
• Create a Heroku account
– Browse to http://signup.heroku.com
• Follow the instructions
– Need not provide a credit card number
– Link your Heroku account to the GitHub
Student Developer Pack
• Procedure unknown

33
 Deploying the App
• Log into Heroku
– Browse to https://id.heroku.com/login
• For Email address enter
rdondero@cs.princeton.edu
• For Password enter yourpassword

34
 Deploying the App
• “Install Heroku” on GitHub
– Inform GitHub that Heroku is permitted to
access the repo
– Browse to
https://github.com/apps/heroku/installations/new
• Click rdondero

35
 Deploying the App
• “Install Heroku” on GitHub (cont.)
– In the resulting Heroku page
• Click Only select repositories
• Click Select repositories
• Choose rdondero/pennyheroku
• Click Save

36
 Deploying the App
• (If necessary) Browse to Heroku
dashboard
– Browse to
https://dashboard.heroku.com/apps
– Click Dashboard

37
 Deploying the App
• Create the app
– In resulting page:
• Click New → Create new app button
– In resulting page:
• For App name enter pennyheroku
• For Choose a region choose United States
• Click Create app button

38
 Deploying the App
• Configure the app
– In resulting page:
• Click GitHub: Connect to GitHub
• For Search for repository to connect to enter
rdondero and pennyheroku
• Click on the Search button
• Click on the Connect button for
rdondero/pennyheroku

39
 Deploying the App
• Configure the app (cont.)
– In the Manual deploys area
• For Choose a branch to deploy select main
• Click the Deploy Branch button
• Observe the build log

40
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

41
 Creating the DB
• In the pennyheroku page
– Click on the Resources tab
• In the resulting Resources page
– Click on the Find more add-ons button
• In the resulting Heroku Add-ons page
– Click on Heroku Postgres

42
 Creating the DB
• In the resulting Heroku Postgres page
– Under Plans & Pricing choose Essential 0
– Click on the Install Heroku Postgres button
• In the resulting Online Order Form page
– As App to provision to choose pennyheroku
– Click on Submit Order Form
• In the resulting pennyheroku page
– Click on Settings tab
– Click on Reveal Config Vars
• Note value of DATABASE_URL
• Save it someplace
• Let’s call it yourdburl

43
 Creating the DB
• Note:
– Heroku reserves the right to change
yourdburl at any time
• Changes DATABASE_URL in Heroku app, but…
• Breaks external apps

44
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

45
 Using the DB
• From a command-line
– Install psql
• Without installing PostgreSQL server
• See
https://www.risingwave.dev/docs/current/install-psq
l-without-postgresql/

46
 Using the DB
• From a command-line (cont.)
– Install psql (Mac)
• First install homebrew; then:

$ brew update
$ brew install libp
$ brew link --force libpq
$

47
 Using the DB
• From a command-line (cont.)
– Install psql (MS Windows)
– Download the installer at
https://www.postgresql.org/download/windows/
– Run the installer
• Select Command Line Tools and uncheck other
options during installation

48
 Using the DB
• From a command-line (cont.)
– Install psql (Linux)

$ sudo apt update

$ sudo apt install postgresql-client
$

49
 Using the DB
Some psql statements

sqlite3 Statement psql Statement

.help \h
.quit \q
.tables \d
.schema table \d table
.read file \i file

50
 Using the DB

$ cat penny.sql
DROP TABLE IF EXISTS books;

CREATE TABLE books (isbn TEXT PRIMARY KEY, author TEXT, title TEXT);

INSERT INTO books (isbn, author, title)

VALUES ('123', 'Kernighan','The Practice of Programming');
INSERT INTO books (isbn, author, title)
VALUES ('234', 'Kernighan','The C Programming Language');
INSERT INTO books (isbn, author, title)
VALUES ('345', 'Sedgewick','Algorithms in C');
$

51
 Using the DB
$ psql yourdburl
yourdbname=> \i penny.sql
DROP TABLE
CREATE TABLE
INSERT 0 1
INSERT 0 1
INSERT 0 1
yourdbname=> SELECT * FROM books;
isbn | author | title
------+-----------+-----------------------------
123 | Kernighan | The Practice of Programming
234 | Kernighan | The C Programming Language
345 | Sedgewick | Algorithms in C
(3 rows)

yourdbname=> \q
$
52
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

53
 Using the DB: Python

Python program

DBAPI
Socket PostgreSQL
psycopg2
PostgreSQL DBMS
protocol
File I/O

Data
store

54
 Using the DB: Python
• Installing the psycopg2 driver:

$ activate333
$ python –m pip install psycopg2
$
or, if that fails:
$ activate333
$ python –m pip install psycopg2-binary
$

55
 Using the DB: Python
• See pennyheroku/database1.py
– Baseline version
– Uses SQLite
$ python database1.py
123
Kernighan
The Practice of Programming

234
Kernighan
The C Programming Language

56
 Using the DB: Python
• See pennyheroku/database2.py
– Uses PostgreSQL
– Uses an environment variable
$ export DATABASE_URL=yourdburl
$ python database2.py
123
Kernighan
The Practice of Programming

234
Kernighan
The C Programming Language

$
57
 Using the DB: Python
• Problem:
– In the context of a Flask app…
– database2.py is too slow

58
 Using the DB: Python
App Server
Concurrent
threads
penny.py

HTTP penny.py
Browser DB
Server
penny.py

penny.py

59
 Using the DB: Python
App Server
Concurrent
threads
penny.py

HTTP penny.py
Browser DB
Server
penny.py

penny.py

penny.py

penny.py

60
 Using the DB: Python
• Solution:
– DB connection pooling
• Maintain a “pool” (queue) of open DB connections
that threads can use

61
 Using the DB: Python
• See pennyrender/database3.py
– Uses DB connection pooling

$ export DATABASE_URL=yourexternaldburl
$ python database3.py
123
Kernighan
The Practice of Programming

234
Kernighan
The C Programming Language

62
 Using the DB: Python
• Problems???
– database3.py requires extra logic for
connection pooling
– database3.py requires maintenance
pgmmers to know SQL
– database3.py is (somewhat) specific to
PostgreSQL

63
 Using the DB: Python
• Solution: SQLAlchemy
– An Object Relational Mapper (ORM) for
Python
• Maps each table to a class
• Maps each row to an object
– See optional lecture and tutorial material for
more thorough description

64
 Using the DB: Python
• Installing SQLAlchemy

$ activate333
$ python -m pip install SQLAlchemy
$

65
 Using the DB: Python
• See pennyheroku/database.py
– Uses SQLAlchemy
$ export DATABASE_URL=yourdburl
$ python database.py
123
Kernighan
The Practice of Programming

234
Kernighan
The C Programming Language

66
 Using the DB: Python
• See pennyheroku/database.py (cont.)

$ export DATABASE_URL=sqlite:///penny.sqlite
$ python database.py
123
Kernighan
The Practice of Programming

234
Kernighan
The C Programming Language

Also works with SQLite on local computer!!!

67
 Agenda
• Deployment options
• Heroku
• Defining the app
• Deploying the app (demo)
• Creating the DB (demo)
• Using the DB (demo)
• Using the DB: Python
• Running the App (demo)

68
 Running the App
• Browse to the app!
– https://pennyheroku-??????.herokuapp.com/

69
 Summary
• The lecture has covered:
– Web app deployment options
– How to deploy a database and web app to
Heroku
• See also:
– Appendix 1: Render vs. Heroku
– Appendix 2: Cloud Service Types

70
 Appendix 1:
Render vs. Heroku

71
 Render vs. Heroku
• Render
– (pro) Easier to create DB
– (pro) Easier to create app
– (pro) Stable DB URL
– (pro) Free
• Heroku
– (pro) Faster deploys
– (pro) Faster initial app launch
– (pro) Faster app???
– (pro) DB does not expire

72
 Render vs. Heroku
• Render
– (con) Slower deploys
– (con) Slower initial app launches
– (con) Slower app???
– (con) DB expires after 1 month
• But easily can move data to new DB
• Heroku
– (con) Harder to create app
– (con) Harder to create DB
– (con) Unstable DB URL
– (con) Costs money
• But non-renewable GitHub Student Developer Pack covers
2 years

73
 Appendix 2:
Cloud Service Types

74
 Cloud Service Types
• Software as a Service (SaaS)
– “The capability provided to the consumer is to
use the provider’s applications running on a
cloud infrastructure.”
-- https://en.wikipedia.org/wiki/Cloud_computing

– Examples: Google Docs, Microsoft Word

Online

75
 Cloud Service Types
• Platform as a Service (PaaS)
– “The capability provided to the consumer is to
deploy onto the cloud infrastructure
consumer-created or acquired applications
created using programming languages, libraries,
services, and tools supported by the provider.”
-- https://en.wikipedia.org/wiki/Cloud_computing

– Examples: Render, Heroku, Google App Engine

76
 Cloud Service Types
• Infrastructure as a Service (IaaS)
– “The capability provided to the consumer is to
provision processing, storage, networks, and
other fundamental computing resources where
the consumer is able to deploy and run arbitrary
software, which can include operating systems
and applications.”
-- https://en.wikipedia.org/wiki/Cloud_computing

– Examples: Amazon Web Services (AWS),

Google Cloud Platform, Microsoft Azure

77
 Cloud Service Types
• Which cloud service type for Penny?
– SaaS: impossible; too narrow
– IaaS: possible, but too broad
– PaaS: just right!

78