IBM Facilitates Serverless Computing With OpenWhisk

“Going serverless” is rapidly becoming the deployment option of choice for developers looking for maximum flexibility.

It just makes sense: Why take on the additional responsibility, complexity, and cost of provisioning and configuring servers, when your code can simply be ready when you are—called on an event basis and billed according to resources consumed, not server time.

What Serverless Actually Means

The name is a little misleading. There are still servers, of course. The difference is that the developer never has to think about them. Someone else, usually a cloud provider, keeps the machines running, patches them and adds capacity when it’s needed.

In a functions-as-a-service (FaaS) model, a developer writes a small piece of code, often a single function, and tells the platform when to run it. That trigger might be an HTTP request, a new file landing in storage, a message on a queue or a scheduled time. The platform starts the code, runs it and shuts it down again. If nothing calls the function, nothing runs and nothing is billed.

That’s a big shift from renting a virtual machine that sits idle most of the day. It also changes how teams plan for traffic. Instead of guessing how many servers they’ll need for a busy period, they let the platform scale the function up and down automatically.

IBM’s Bet on OpenWhisk

IBM has made significant investments in OpenWhisk, an open-source, event-driven FaaS platform. OpenWhisk is now generally available via Bluemix, IBM’s cloud platform.

Recently accepted into the open community of Apache developers, OpenWhisk now includes new services such as:

• Instant debugging for Node.js, Python and Swift actions
• Integration with Message Hub, a Bluemix-hosted Apache Kafka service for real-time build outs of data pipelines and streaming apps
• Support for new runtimes such as Java, Node v6, Python and Swift v3
• An extension for Visual Studio Code
• A new and improved user interface to simplify browser-based development and testing

Making OpenWhisk open source set it apart from some rivals. Amazon’s AWS Lambda, which helped popularize the serverless idea, is a proprietary service that only runs on Amazon’s cloud. Because OpenWhisk’s code is public, companies can run it on their own hardware or on another provider’s infrastructure, not just on IBM’s.

Santander’s Real-Time Banking Plans

Santander Group, one of the world’s top banks by market capitalization, is adopting OpenWhisk to speed time-to-market for its digital banking services.

“OpenWhisk provides the instant infrastructure we need for intense tasks and unexpected peaks in workload,” said Luis Enriquez, head of platform engineering and architecture at Santander Group, “and is a key building block as we move to a real-time and event-driven architecture.”

Banking is a good fit for this model. Customer activity comes in bursts, around paydays, at the end of the month or when a new app feature goes live, and many tasks are naturally triggered by events such as a payment or a login. Paying only for the computing that those events actually use can be cheaper than keeping enough servers on standby for the busiest moments.

Learn more about why OpenWhisk is the most flexible serverless computing platform for enterprise here.

What Happened to OpenWhisk Next

OpenWhisk’s time in the Apache Incubator, where new projects prove they can run as independent open-source communities, started on November 23, 2016. The project graduated on July 17, 2019, becoming a full top-level Apache project. That meant it was no longer seen as an IBM project with outside helpers, but as a community-governed one in its own right.

On IBM’s side, the hosted service was rebranded. By August 2017, Bluemix OpenWhisk was being offered as IBM Cloud Functions, and IBM went on to retire the Bluemix brand, bringing its services under the single name IBM Cloud.

The open-source route also brought in other big users. Adobe’s own serverless platform, Adobe I/O Runtime, is built on top of Apache OpenWhisk, which is exactly the kind of reuse an open project makes possible.

IBM itself has since moved on. Its newer serverless service, IBM Cloud Code Engine, runs apps, batch jobs and functions in one place, and IBM has published guidance on moving workloads from IBM Cloud Functions to Code Engine. Serverless functions are still part of IBM’s cloud, just no longer under the OpenWhisk name.

What It Means for Developers

The broad idea IBM was backing in 2016 has held up. Event-driven functions are now a standard option on every major cloud, and developers routinely use them for webhooks, data processing, chatbots and other short, bursty jobs.

The OpenWhisk story also shows why it’s worth checking how portable a platform is before building on it. Brand names, consoles and hosted services change, as IBM’s did several times in a few years. Code written against an open standard or open-source runtime is easier to move when that happens, whether to another provider or to a company’s own servers.