Functions as a Service
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
OpenDev Sysadmins 20a8baf72d OpenDev Migration Patch 3 years ago
devstack Replace git:// URLs with https:// 3 years ago
docs docs: fix more broken links 5 years ago
examples Replace git:// URLs with https:// 3 years ago
migrations New project name: Picassa 5 years ago
picasso Issue #13: Improve logging in controllers 5 years ago
scripts New project name: Picassa 5 years ago
service Cleanup controllers 5 years ago
.gitignore Issue #17: Introduce Bandit static security checks 5 years ago
.gitreview OpenDev Migration Patch 3 years ago
Dockerfile New project name: Picassa 5 years ago
Dockerfile.env.example New project name: Picassa 5 years ago
LICENSE update copyright on apache license 5 years ago Initial commit 5 years ago Add topic related to private and public functions 5 years ago uppercase dot md 5 years ago
alembic.ini Simplify migrations automation 5 years ago
requirements.txt Use newer version of python-functionsclient v0.0.3 5 years ago
setup.cfg Fixing links from iron-io to openstack GitHub orgs 5 years ago Use PBR for 5 years ago
test-requirements.txt Update test requirement 5 years ago
tox.ini docs update (#32) 5 years ago


Functions-as-a-Service (FaaS) on OpenStack


Picasso aims to provide an API abstraction layer for Functions-as-a-Service (FaaS) on OpenStack.

What is FaaS?

Functions as a service is a new paradigm in computing that enables simplicity, efficiency and scalability for both developers and operators. It's important to distinguish the two, because the benefits differ:

Benefits for developers

The main benefits that most people refer to are on the developer side and they include:

  • No servers to manage -- you just upload your code and the platform deals with the infrastructure
  • Super simple coding -- no more monoliths! Just simple little bits of code
  • Pay by the milliseconds your code is executing -- unlike a typical application that runs 24/7, and you're paying 24/7, functions only run when needed

Since you'll be running IronFunctions yourself, the paying part may not apply, but it does apply to cost savings on your infrastructure bills as you'll read below.

Benefits for operators

If you will be operating IronFunctions, then the benefits are different, but related.

  • Extremely efficient use of resources
    • Unlike an app/API/microservice that consumes resources 24/7 whether they are in use or not, functions are time sliced across your infrastructure and only consume resources while they are actually doing something
  • Easy to manage and scale
    • Single system for code written in any language or any technology
    • Single system to monitor
    • Scaling is the same for all functions, you don't scale each app independently
    • Scaling is simply adding more IronFunctions nodes

System requirements

  • Operating system: Linux/MacOS
  • Python version: 3.5 or greater
  • Database: MySQL 5.7 or greater

Quick-start guide

Create a Python3.5 virtualenv

$ virtualenv -p python3.5 .venv
$ source .venv/bin/activate

Install Picasso dependencies

$ pip install -r requirements.txt -r test-requirements.txt

Install Picasso

$ pip install -e .

Install MySQL if you haven't already, and create a new database for functions

$ mysql -uroot -p -e "CREATE DATABASE functions"


Once all dependencies are installed it is necessary to run database migrations. First, set the following environment variable:

export PICASSO_MIGRATIONS_DB=mysql+pymysql://root:root@localhost/functions

Then use alembic to apply the migrations

$ alembic upgrade head

Starting the Picasso API server

$ picasso-api --help

Usage: picasso-api [OPTIONS]

  Starts Picasso API service

  --host TEXT                    API service bind host.
  --port INTEGER                 API service bind port.
  --db-uri TEXT                  Picasso persistence storage URI.
  --keystone-endpoint TEXT       OpenStack Identity service endpoint.
  --functions-url TEXT           IronFunctions API URL
  --log-level TEXT               Logging file
  --log-file TEXT                Log file path
  --help                         Show this message and exit.

The following are the minimum required options to start the Picasso API service:

 --db-uri mysql://root:root@
 --log-level INFO

Building and Running Picasso in Docker

Install Docker engine if you haven't already.

From the Picasso repo, build a Docker image using the following commands:

export DOCKER_HOST=tcp://<docker-host>:<docker-port>
docker build -t picasso-api -f Dockerfile .

To start the container, pass in the required env vars with --env-file or by entering all required values in -e <KEY>=<VALUE> format to the docker run command.

Example Dockerfile.env

docker run -d -p 10001:10001 --env-file Dockerfile.env picasso-api

Once the container is started, check if the service in running. In your web browser navigate to:


or using the CLI:

curl -X GET http://<docker-host>:10001/api/swagger.json | python -mjson.tool

API docs

API docs are discoverable via Swagger. Just launch the Picasso API and browse to:


Picasso provides private and public functions

Using the Picasso API, it is possible to create two type of functions - private and public.

  • Private functions are defined as functions that belong to a specific OpenStack project, and the execution of the function requires passing in authorization headers.
  • Public functions belong to a specific OpenStack project ID, but do not require any authorization headers to execute, allowing the function to be shared with anyone.

Private and public functions are handled by different URL handlers:

<picasso-api-host>:<picasso-api-port>/r/{project}/{app}/{route} for private functions
<picasso-api-host>:<picasso-api-port>/r/{app}/{route} for public functions

See OpenStack alarming example for more details.

Testing Picasso



Join us on Slack!