By Richard Thomchick, VMware Product Manager, Developer Portal
The Cloud Foundry Summit took place on June 13th through 15th in Santa Clara, Calif. Billed as a premier event in Silicon Valley for enterprise applications developers, the summit focused largely on streamlining the software development pipeline, adopting microservices, and continuous delivery.
The keynotes highlighted enterprises that are solving problems by leveraging open source and deploying Cloud Foundry. A keynote presenter was Kyla McMullen, Ph.D., an assistant professor at the University of Florida. She runs the SoundPadLab (http://kylamcmullen.com/yunhaow/) at the University of Florida, and her keynote discussed how her team uses open source to create 3D (“spatial”) audio to make assistive technology, build accessible interfaces, and enhance big data.
McMullen said that her research often brings her and her team in contact with people who have physical impairments, and her keynote emphasized the need to engage diverse user bases in the systems design process. Using the example of an automated soap dispenser that could not detect dark-skinned hands, McMullen pointed out that the assumptions we make in software design can impact the real world. Her view is that the internet of things (IoT) and artificial intelligence (AI) technology will magnify these effects, and that open source will be an even greater imperative because it allows for the introspection of the code so that such assumptions can be identified.
Another compelling presentation was by Alex Sologub of Altoros on the potential pitfalls of adopting micro-services to re-architect monolithic legacy applications. The basic problem, as Sologub put it, is ignoring the data. When teams attempt to “break down the monolith” too early, without analyzing the user experience, performance metrics, use cases, and source code, they often end up with more apps to test, deploy, and support; more development and infrastructure overhead; unnecessary complexity; wasted resources; and backlogs that are dominated by technology.
Linda Nichols of Emerging Technology Advisors explained how to create a serverless cat detection system with IoT and image-recognition APIs. This session was a DIY-oriented demonstration of how to turn a simple motion sensor into a smart “serverless” system that alerts users when specific objects (in this case, cats) are detected.
Using the IBM Watson IoT Platform and Node.js, Nichols connected a Raspberry Pi with IBM Bluemix OpenWhisk functions and IBM Watson image recognition APIs. For her demo, Nichols used a Raspberry Pi board and camera module with the Johnny-Five platform (http://johnny-five.io) installed, which she connected to IBM Watson image-recognition APIs. The system was able to detect cats and ignore other categories of moving objects, and sent a text alert to Nichols when it detected (a photograph of) a cat. Nichols also used “serverless” IBM Bluemix OpenWhisk functions to reduce the complexity of the system architecture and the amount of static code stored on the Raspberry Pi.
Nichols also points out that other technologies (e.g., Arduino, Swift, Nest, Google Image Recognition APIs) could just as easily have been utilized to create this type of solution, a side remark that reveals just how quickly IoT services have emerged over the last 12 to 18 months.
Another highlight of the conference was a talk titled “A Smooth Path to Cloud Foundry on Containers,” by Jan von Löwenstein of SAP. He is a senior developer on the BOSH OpenStack CPI project, and he talked about the work his team is doing to enable BOSH deployment using a native Kubernetes CPI. The purpose of the project (https://github.com/SAP/bosh-kubernetes-cpi-release) is to provide a fully functional BOSH CPI that allows for deployment of any BOSH release to any Kubernetes cluster, with the ultimate goal of making container platforms just another deployment choice for Cloud Foundry. The code essentially maps Kubernetes concepts to BOSH—for example, a Pod with a single container is used to represent a BOSH virtual machine, and persistent volume claims are used to represent BOSH persistent disks.
Löwenstein noted that there are issues he has not been able to resolve between older and newer versions with different feature support. First, the tool can only be used with older versions of Minikube and Kubernetes currently, while problems with newer versions are being addressed. Second, dynamic networks are supported at this time, so releases that require static IPs are not supported. Third, the BOSH agent must run in privileged containers, and SSL verification is not yet supported to ensure secure communications with Kubernetes.
With a growing community, Cloud Foundry is clearly a project with feature enhancements in high demand as enterprises modernize applications and tooling. The event highlighted many use cases across existing and new technologies, and brought attention to the need to deliver better user experience through diversity and open ecosystems. Cloud Foundry Summit was an event that brought together technologies, communities, people, best practices, and lessons learned. The sessions have now been shared on the Cloud Foundry You-Tube channel and can be found here.
Follow the Cloud-Native Applications blog and Twitter @CloudNativeApps to learn about more cloud-native related news.