Compute Services

Exposing OpenWhisk RESTful APIs with API Gateway

Share this post:

openwhisk-hero-image API GatewayWith the recent promotion of OpenWhisk to general availability, you not only get the incredibly powerful, scalable, event-driven programming model for building your actions (functions) using serverless computing, but something else slipped into the release that you may not have noticed:

The OpenWhisk API Gateway

The API gateway is an experimental new feature that enables you to easily expose your OpenWhisk actions as RESTful endpoints. You can assign actions to specific endpoints, and even have verbs (get, put, post, delete) from the same endpoint assigned to different actions. You can already consume OpenWhisk actions from other applications using the JavaScript or iOS/Swift client SDKs, and now the API gateway gives you an easy path to expose your OpenWhisk actions to other non-credentialed applications.

There will be two different approaches you can to expose your actions with the API gateway:

  • Assigning API endpoint/verb combinations to specific actions individually
  • Using a Swagger config file to map API endpoints to actions (currently still in development)

Note: This is different from the ability to expose your OpenWhisk actions as managed APIs using API Connect.

A simple action: Fibonacci generator

Next we will examine how to expose an action as a REST API, but first let’s take a look at the action itself. I’m starting with a fibonacci generator. It’s a relatively simple action that generates numbers in a fibonacci sequence, where every number after the first two is the sum of the two preceding values. When invoking this action from the command line, you specify a num parameter (for the Nth place in the sequence), and it will return the value, the complete sequence, and the number of recursive invocations of the fibonacci method:

$ wsk action invoke fibonacci -p num 5 -b -r
   "n": 5,
   "value": 8,
   "sequence": [1, 1, 2, 3, 5, 8],
   "invocations": 9

You can download it and check it out for yourself: fibonacci.js

Mapping actions to endpoints

Now, let’s examine how a specific action can be associated with an API endpoint/verb. From the OpenWhisk CLI, you must specify a base path, the api path (path component for the action), the verb (get, post, put, delete), and then the action.  See more variations in the docs.

In this case, the base path is is /api-demo/v1, the api path is /fibonacci, the verb is get, and the action is fibonacci:

$ wsk api-experimental create /api-demo/v1 /fibonacci get fibonacci
ok: created API /api-demo/v1/fibonacci GET for action /_/fibonacci

This exposed my fibonacci action as a get request for the URL shown directly above. You can check it out with different fibonacci values here:

Parameters that are passed via the query string will be available in the args object passed into the action’s main function.  For example, a query string parameter ?num=10 will be passed into your OpenWhisk args as:


View the fibonacci links above to see how I am passing data into this API endpoint.

Next Steps

If you haven’t already, be sure to download the latest command line interface. For additional information, be sure to check out the docs within the OpenWhisk project’s repo, and you can always run wsk api-experimental —help for ore detail in the CLI. Keep in mind, this is an early, experimental feature that provides users an early opportunity to try it out and provide feedback. We’ve already received several enhancement requests, but definitely let the team know if there are other things you’d like to see added.

More Compute Services stories

What the stats say about container development

59% improved application quality and reduced defects. 57% reduced application downtime and costs. All adopted container development.   In 2017, IBM conducted an in-depth research study on the state of container adoption across all industries, startups to enterprises. The study reveals the most important solutions driving usage and highlights the key challenges that must be addressed by cloud providers. […]

Continue reading

Process large data sets at massive scale with PyWren over IBM Cloud Functions

(Ed.–Josep Sampé–Universitat Rovira i Virgili–co-authored this post.) Let’s say you write a function in Python to process and analyze some data. You successfully test the function using a small amount of data and now you want to run the function as a serverless action at massive scale, with parallelism, against terabytes of data. What options do you have? Obviously, […]

Continue reading

Combine organizational transformation and IBM Cloud infrastructure for a successful AI journey

With our recent cloud infrastructure and Deep-Learning-as-a-Service (DLaaS) announcements, IBM Cloud is a key contributor to the push towards AI. We’ve delivered a comprehensive suite of AI tools, high performance bare metal servers, and NVIDIA® GPUs that enables companies of all sizes to analyze complex unstructured data faster, more thoroughly and accurately, and at a far less cost than ever before.

Continue reading