Tony from Linux Recruit had the great idea of hosting a DevOps Exchange meetup at Vungle while he was in town for Kubecon, and it was always our plan to eventually take what has become the world’s largest DevOps meetup in London, over to the states, and see how it went over there.
Suffice to say, starting at ground zero, this was much smaller than our usual 150-200 attendee meetups back in London, but it was an awesome start and once I’m fully settled over in San Francisco I plan to host more of them with more great speakers. Thanks a lot to our first attendees for coming, it was great to meet the San Francisco DevOps community, and I look forward to meeting you all again at future meetups!
You can join DevOps Exchange San Francisco here to get updates on when the next meetup will be. Until then, enjoy the videos from our first meetup!
Ian Lewis (Google)
Managing Cross-Team Resources at Scale
Getting the most out of resources you are paying for can be a challenge. Virtualization allowed us to create clusters of machines to run parallel workloads easily. But, traditionally, organizations only got about 40% utilization out of those clusters. Containerizing those workloads and running them with a scheduler like Kubernetes can help you get better utilization by running different kinds of workloads together on the same cluster. Even still, you will want some control over how many cluster resources each of your teams or applications use. In this talk I will help explain how to run workloads from multiple teams in the same cluster and limit those teams to a certain amount of resources. I will look at what happens when clusters run out of resources, and walk through how to debug and fix allocation issues. As an added bonus, I will also talk about how applications and services can operate across team boundaries.
Joseph Schneider (Bunchball)
Ephemeral Namespaces: Full stack testing with Kubernetes
Switching to a new paradigm doesn’t happen overnight. For our transition to Kubernetes, Bunchball is combining the power of Jenkins and Kubernetes namespaces to enable one click deploys of our entire technology stack on demand in minutes for full regression tests. It’s a single command to tear it back down as well when we’re done. It’s a great smoke test of our Kubernetes cluster as well.
Colin Fallwell (AppDynamics)
Performance Engineering the SDLC
Containers create new challenges as well as opportunities around managing application performance across the SDLC. Building effective CI/CD pipelines with robust gates are essential to providing the tight feedback loops necessary for high Net Promoter Scores. This session will focus on how to leverage Kubernetes, performance engineering best practices and the right mix of metrics across the SDLC to supercharge developer productivity and create organization-wide innovation potential.