I had the honor of introducing Magnum during the morning Keynote of the OpenStack Summit today before an audience of 6,000 attendees. You can tell these are authentic live demos because there were some tiny quirks. If you can take the 50 minutes to watch the whole thing, It’s worth the time. If you want to fast forward to the parts with Magnum and Kubernetes, check out these sections:
21:36-31:46 Magnum Introduction and Demo
42:31-50:00 Google+Racksapce Multi-Cloud Demo with Kubernetes
Note that toward the end of the Rackspace + Google demo, Sandeep refreshes the page, and an unexpected error happened because the load balancer continued routing requests to the application server with the database he turned off. This happened because it had not yet drained the active connections. We had interference from the audience!! They viewed the IP Address of the app, and accessed it while Sandeep was showing the demo. The Rackspace load balancer does not close new connections to a node in the middle of servicing active concurrent connections unless you *remove* the node. In other words, disabling a node is non-disruptive. If there were no other requests happening at the time we were showing this, it would have acted like it did the night before in the hotel.
If Sandeep had pressed refresh one more time before giving up, it would have worked! Members of the audience were visiting the page literally two seconds later, and they said the app remained on-line flawlessly for another 300+ hits while we were still on stage, and after we finished. They showed us backstage afterward that it was actually still up, and responding correctly.
I am happy that our Kubernetes and OpenStack communities are working together to make sure that container technology is equally useful and accessible for all types of clouds.
Today OpenSource.com published an interview with me about an an upcoming talk I’m giving in Vancouver next week about applying proven best practices to open source software development. Community open source development requires a different approach than Agile, where unity of distributed contributors matters just as much as team unity on a local development team.
Magnum is a project that brings application containers, including Docker, to OpenStack as a first class cloud resource. It allows cloud operators to set up and offer Containers-as-a-Service on their clouds, leveraging tools such as Kubernetes and Docker Swarm with a modular approach that allows for future proofing of your container strategy. Magnum has diverse community support from 18 different participating companies (and growing!), many of which have allocated full time engineers to work exclusively on the project.
Solum is a project aimed at simplifying automation of the software development lifecycle, and the management of developed applications. Developers using Solum connected to a cloud can check in their code to a Git source code repository, and Solum will automatically build and test the software, and stage it for automated deployment (CI/CD). It produces Docker Container images for hosting your apps that are portable to any infrastructure that can run a modern Linux kernel.
The article touches on best practices I learned from leading these collaborative open source development efforts using the OpenStack development style, and adding my own slight twists. To my amazement, many of the things I tried as deviations from OpenStack’s way of doing things did not work well, and I ended up simply adopting prevailing best practices as I systematically proved each. The OpenStack approach was adapted from the best of other successful communities, such as Ubuntu. If you are attending the OpenStack Summit in Vancouver next week, I look forward to having you in my session!
Today Timothy Prickett Morgan from The Platform published a good article in about our work on Magnum. Over 40 engineers from 18 different affiliations have contributed to Magnum since we began in November 2014. Next week we will be showing Magnum to the 5000+ attendees of the OpenStack Summit in Vancouver, BC.
Magnum is a Platform-as-a-Service solution for OpenStack. We joined as an official OpenStack project on March 24th 2015. Magnum combines OpenStack with a Container Orchestration Engine of your choice (Kubernetes or Docker Swarm so far) to provide a vertically integrated container service that your OpenStack cloud users can enjoy with the same cloud account and credentials they use to provision other cloud resources, such as Compute Storage, Networks, etc. It stands apart from other container software solutions because it’s implemented as Multi-Tenant from the bottom up, so you can have different users and project groups who are properly isolated from each other so that have no visibility into or access to other containers in the system.
Magnum integrates with OpenStack Identity (Keystone), OpenStack Orchestration (Heat), OpenStack Images (Glance), and OpenStack Networking (Neutron). Using Heat allows us to create Nova instances for to run containers on.
Cloud operators are struggling to figure out what container software they should be working with, and how they should offer it to their users. OpenStack Magnum helps simplify that choice by making a selection of the leading software available, and a modular interface to allow different Container Orchestration Engines to be plugged in later, if the prevailing favorite changes over time. OpenStack is a completely open community not driven by any single sponsor. That gives cloud operators a real sense of confidence that no change in control, or change in direction of any single software vendor will put OpenStack at risk. Pretty awesome!
If you are planning to attend the OpenStack Summit, you can learn more about Magnum at our session. I look forward to seeing you in the audience.