Pivotal announced the latest update to its Pivotal Cloud Foundry platform on Dec. 5, along with a preview of the company's Kubernetes-powered container service and the launch of a new serverless effort.
Pivotal Cloud Foundry (PCF), Pivotal's core platform since the company was spun out from VMware in April 2013, is based on the open-source Cloud Foundry platform-as-a-service (PaaS) project. The PaaS market has shifted over the last four years and so has Pivotal, which is why the new Kubernetes and serverless projects are important to the company and its customers.
"We're adding new first-class entities to Pivotal Cloud Foundry in response to evolving market demand," James Watters, senior vice president of strategy at Pivotal. "We're in a great position to expand with other meaningful interfaces for our clients because we've had a lot of success with our core platform."
The core PCF platform is based on the Cloud Foundry project, which is updated approximately once a week. In contrast, PCF is updated once a quarter, providing an enterprise-ready distribution, Watters said.
Going beyond the core PaaS functionality provided by Cloud Foundry, Pivotal now also supports running containers as a service with the Pivotal Container Service (PKS), it uses the open-source Kubernetes container orchestration system along with Cloud Foundry's BOSH life cycle management capabilities.
Pivotal Function Service
While PaaS and containers both are now established market categories, an emerging area of IT is serverless computing, which was first popularized by the Amazon Web Services Lambda service. Pivotal is calling its new serverless offering the Pivotal Function Service (PFS) to help differentiate it and articulate what the service actually does.
There is a lot of confusion in the market today about what the term "serverless" means, according to Watters. "Serverless is really any time you're not managing the servers and the operating system, and you're just using a programming interface," he said.
In contrast, PFS is event-based, Watters said. It starts when an event is triggered by a message queue or other programmatic trigger and will scale up automatically based on the event and then will scale down and turn off automatically when the event is completed. An example of where PFS can help is with financial services clients that only need to run a certain procedure at the end of the day and don't need to have a long running, always-on server, he said.
PFS uses short-lived containers that are rapidly scaled up and then shut down to provide the event-driven service. The service connects the function, which runs inside a container, to a given event, Watters said.
For developers looking to build functions for PFS, Pivotal has a technology called Spring Cloud Function that enables developers to automatically wire a data type into an event that will execute and then expire. Spring Cloud Function also includes a function registry that allows developers to build and share PFS functions.
Cloud Foundry vs. Kubernetes
Cloud Foundry has often been labeled a PaaS, while Kubernetes provides a container-as-a-service capability. Understanding when to use one approach over the other isn't always an easy task, but it's one that Pivotal is actively talking to its customers about.
By providing multiple options, Pivotal is working to enable different use cases, according to Watters. He added that the different approaches to application deployment are all about different levels of abstraction.
"If you have something that runs once a day and then expires, a function service is great for that," Watters said. "If you have a traditional Java application, then the Cloud Foundry approach is the right way to go."
The Pivotal Cloud Foundry platform provides robust management and control services for applications, Watters said. In contrast, he noted that the container service is really useful for things that are not Java applications, but rather are more elastic in nature.
"We have a whole consulting framework where we help customers pick the right abstraction for efficiency," Watters said. "We don't want to be biased toward one abstraction or another; we just want to help provide the right fit for the workload."
Although Pivotal Cloud Foundry is often called a PaaS, Watters said he prefers to think of it in its current iteration as a cloud-native platform that has different programming interface abstractions. "Sometimes people get caught up in the rivalries between the abstractions, versus thinking about what the best fit is for a given use case," he said.