KubeCon 2019 wrap up

KubeCon 2019 wrap up


 KubeCon + CloudNative North America 2019 Conference (KubeCon) just ended. I wanted to jot down some impressions while they were fresh in my mind.

The talks I found most interesting were the case studies on real-world experience adopting Kubernetes and entering production by some of the most well-known companies in the world. In the upper right of
the Program Schedule, you can filter the talks to get Case Studies, Maintainer Track or say Keynotes. The best talk I saw was the case study from Airbnb and was titled 10 Weird Ways to Blow Up Your Kubernetes. The talk struck the right balance of humor and problem analysis and solutions used to move Kubernetes into production in their environment. Melanie Cebula and Bruce Sherrod delivered the talk to a standing room only audience in one of the larger conference rooms at the venue. As for all talks, the slides and the recording of the session are available. I attended another talk by Airbnb and a retrospective talk on Kubernetes adoption by Tinder. The talk by Tinder was one of the last of the conference, and the large room was again packed.

Established web scale giants talked primarily about moving existing Cloud Native applications into a

Kubernetes framework to gain mobility for their applications amongst different hosting environments, and accelerate their DevOps practices. The application cluster sizes they described numbered in the thousands, and scaling in Kubernetes was a challenge for these companies as they gained expertise in the platform. One refrain heard again and again in the talks I attended was that change is the one constant in the Kubernetes world. Unfortunately, for example, this might mean something that worked before Kubernetes 1.12 may stop working on a new release. The Kubernetes platform is definitely still maturing, but this is not hindering its adoption. As features are cleaned up the semantics of some fields might change and break your existing Kubernetes application, but the fix is usually found quickly. What’s great about events like KubeCon is that the user community is openly sharing experience and evolving best practices so that others can gain from their learnings.

The informal Meet the Maintainer “pavilion” sessions were also cool. They allowed you to get one-on-one time with people working on the core technologies. These were simply spaced tables in the vendor show areas where the maintainers sat and made themselves available to answer questions. I eavesdropped on a couple of sessions to see what people were worrying about.

 After visiting with a couple of the Rook maintainers at the pavilion, I went to listen to Rook: Cloud-Native Storage Orchestration (Introduction and Deep Dive). Rook looks to provide a standard API for storage operators for Kubernetes. The technical deep dive session was 90 minutes long. One interesting point was when the opening speaker asked: “How many people have first heard about Rook this week at the conference?” At a glance, I would say fully half the room of a couple of hundred people raised their hands. I found this fascinating. KubeCon is the one place where people go to get up to speed on all things Kubernetes. And there were a lot of newcomers at KubeCon using the conference to learn more about the technology and the players.

As I mentioned above, CNCF has a standard process for project adoption. In a bit of introspection, the Rook maintainers described their timeline of proceeding through the steps of Sandbox→ Early Adopters → Graduation. A very practical result from a core technology talk. The second half of the talk described some of the specific storage technologies that integrate into Kubernetes through the Rook framework. There’s a lot more work to be done on Rook and supported storage options.

In a demonstration of inclusion, there were several talks about container orchestration on Windows. I found the Introduction to Windows Containers in Kubernetes interesting in that it illustrates the type of work coming out of the Windows Special Interest Group. Participation in the SIGs is as simple as showing up. The list of active SIGs and working groups can always be found online.

KubeCon this year drew more than 12,000 attendees, up from 8,000 last year, to the San Diego Convention Center. The conference hotels filled up and then some, and San Diego had some much-needed rain one night. On the importance of this conference and acknowledgment of where it came from, Stu Miniman said on-site in theCUBE introduction“If it wasn’t for Docker we wouldn’t be in a 12,500 people […] event this week. … When the orchestration layer came in … Kubernetes sucked the air out of the Cloud Native world.” A remarkable evolution in less than five years.

The Cloud Native Computing Foundation (CNCF) and Kubernetes is supported by a broad set of companies and technologies, and the breadth of the problems being tackled by the ecosystem is amazing. CNCF has very rapidly established the framework to evolve the Kubernetes ecosystem in a straightforward and inclusive manner.

The energy level was high as people moved from session to session. There were more than 300 presentations, with 20 or more presentations running in parallel, making it impossible to see everything you wanted to. But no fear, the sessions were recorded (sponsored by Google Cloud) and are already available online. Like the 2018 presentations, these recorded talks are a great resource to understand the current state of the Kubernetes ecosystem.

The size of the industry investment into Kubernetes was apparent from the more than 200 sponsors with heavily trafficked booths. More than 70 of these vendors were designated “start-ups”, and that included DriveScale. Allowing for the emerging players in the Kubernetes and Cloud Native world made the vendor show floor pretty cool.

We found that the visitors to our booth were extremely interested in our solution for Kubernetes and focused on gaining knowledge about how to successfully deploy their applications on Kubernetes. DriveScale has a lot of follow up after the show, and from our perspective the show was successful. Also, many of the companies sponsoring KubeCon mentioned they were hiring, via job boards on-site or by quick mentions during their talks.

All in all, I had a great time. One of the coolest things I learned at the conference was how to pull up and share my QR code on LinkedIn. This turned out to be pretty important as I connected with a lot of the attendees I came across as I went from session to session or visited with the vendors on the show floor.  I shared the QR Code trick that Gavin showed me with a lot of other attendees. Really, this was easily the best conference to network at I’ve been to.

The second coolest thing was the semi-custom t-shirt machine at D2iQ’s booth.  Who is D2iQ you may ask? The answer is interesting on two levels.

First, the company D2iQ was formerly known as Mesosphere, a pioneer in developing technology to orchestrate application clusters and containers. The history of value that virtualization provided ensured that containerization was rapidly accepted as a given in the future of the data center. Where the real problem lay was that containerization encouraged microservices and massive scale-out and it quickly became apparent that orchestration was the great unsolved problem for containers. Apache Mesos was one of the early “container” orchestration tools that emerged, before Kubernetes “sucked the air out of the room”, with regards to orchestration. In a clever bit of rapid response and rebranding, Mesosphere added Kubernetes orchestration to their Marathon platform and became D2iQ.

The “D2” part of the name means Day 2 in DevOps parlance. Day 1 is roughly adoption and pilot of a new technology, Day 2 is deriving benefit from the technology for your organization in the large, that is, going into production. D2iQ, as does RedHat and VMware, wants to be the trusted partner to established enterprises in rolling out Kubernetes in their data center or the cloud. Given the number of people new to Kubernetes at the conference, D2iQ’s approach is probably a great one. I finally got the meaning of the semi-custom t-shirt operation they had going on in their booth in light of their strategy. The line was way too long for me to wait and get a t-shirt.

Wednesday I had the opportunity to sit down with Michael Waite, from our technology partner RedHat to talk about the current state of affairs with Kubernetes and DriveScale’s integration with Kubernetes to support containerized data intensive applications. We had a great chat and some random tangents which may or may not get cut out of the podcast. Check back in the next week or two for the podcast.

I had a great time at KubeCon. It was an incredibly well run and organized conference from check-in to departure. I have never seen so many coffee stations so conveniently located at a conference before in my life. This attention to seemingly small details and larger ones like rapidly getting the session recordings online soon after the talk was finished speaks volumes to the dedication and capabilities of Angela Brown and her global events team at CNCF and the Linux Foundation. I’m looking forward to their conference transparency report for this year’s KubeCon. The 2018 report is chock full of statistics to gain insights on Kubernetes ecosystem adoption and maturity.

I plan on submitting a talk for the KubeCon + CloudNativeCon Europe 2020in Amsterdam running from March 30 – April 2. Wish me luck on acceptance!

About the Author:

Brian Pawlowski has years of experience in building technologies and leading teams in high-growth environments at global technology companies. He is currently the CTO at DriveScale. Previously, Brian served as Vice President and Chief Architect at Pure Storage, focused on improving the user experience for the all-flash storage platform provider’s rapidly growing customer base. As CTO at storage pioneer NetApp, Brian led the first SAN product effort, founded NetApp labs to collaborate with universities and research centers on new directions in data center architectures.

Leave A Comment