Now that IBM and Adobe Systems are jointly leading the charge in terms of promoting adoption of an open-source serverless computing framework known as OpenWhisk, the two companies have implemented an application programming interface (API) gateway to make OpenWhisk more accessible to applications. Next up, the two companies plan to embed an instance of Kubernetes container orchestration engine within OpenWhisk to enable the framework to be even more efficient.
Michael Behrendt, chief architect for serverless, functional-aaS and event-driven architectures at IBM, says as OpenWhisk continues to evolve IBM expects many stateless applications developed using Docker to migrate to an event-driven architecture that is better-suited for processing stateless microservices, which generally make use of REST APIs to access infrastructure resources. At the same time, however, he says IBM also expects to see a sharp increase in the number of stateful applications being deployed using containers. Most of those applications will be deployed on the same cloud as stateless applications to reduce overall latency. But as a rule, stateful applications will not lend themselves as well to serverless computing frameworks.
At present, IBM has a single instance of OpenWhisk deployed on a Bluemix platform-as-a-service (PaaS) environment hosted in a data center in Dallas. Behrendt says IBM is working toward deploying OpenWhisk more widely as a cloud service, while Adobe is planning to make use of OpenWhisk to help drive its own Adobe.io cloud service.
Behrendt says OpenWhisk is differentiated from rival serverless computing frameworks such as Amazon Web Services’ (AWS) Lambda or Google Cloud Functions because it’s an open-source framework, so IT organizations can move between instances of OpenWhisk at will.
OpenWhisk also supports a much broader range of programming languages, including Java, Python, Node.js, Swift or any language that can be used to create a Docker container. In addition, he says, OpenWhisk provides IT organizations with more granular controls over the amount of memory an application can consume, the amount of time it can run, and access to the logs OpenWhisk generates.
In general, Behrendt says developers will need to employ both serverless frameworks and Docker containers. In both cases, use of IT infrastructure inside and out of the cloud will become substantially more efficient. Those costs savings, in turn, should be passed on to customers either in the form of lower cloud prices or more applications being deployed on servers they acquire and run themselves. Obviously, providers of those platforms are hoping lower costs will lead to more overall consumption. But some providers of those platforms may find it too challenging to survive. That, in turn, could serve as a catalyst for even more consolidation.
Naturally, It may take a while before serverless computing frameworks becomes a mainstream IT architecture. But it’s already clear they now are one of the fastest growing services in the cloud. The next issue facing IT operations teams now will be finding a way to centrally manage multiple types of computing frameworks.