The conference is located in a cool and sort of strange place for a conference, out kind of in the lake, called the Navy Pier. Honestly, I dig it, it’s a cool place for a conference. I enjoyed the ~15 minute walk from the hotel to the location too, because it’s right there on the tip of the pier, as shown in the fancy map below.
The workshop is going well. Bridget is filling up student’s brains and I’m going to dork around Kuberneting some Cassandra and Node.js for my talk. I’m pretty stoked as the talk has given me a good excuse to delve into some Node.js again, from a nodal systemic viewpoint, “Node Systems for Node.js Services on Nodes of Systemic Nodal Systems” this Thursday.
KubeCon 2017 in Austin was great! Informative, entertaining, and overall a giant win for me and a whole lot of other people. KubeCon has officially entered into my “Will Return To Conferences” list. Austin also helped, as it’s a great city for tech conferences. It has a few issues of course, but overall it’s a spectacular city for technology conferences, rating much higher than Las Vegas, San Francisco, and a whole slew of other places in my book.
If you’re experienced, interested in getting involved with open source, or involved with site reliability and operations side of technology, KubeCon has become one of the preeminent conferences to attend. There are workshops, training, impromptu discussions, learning moments, presentations, and more. In addition the sponsor booths and showcase hall was hands down one of the best I’ve seen in more than a year or two. Music, food, and a good comfortable environment to meet, discuss, or just hack on new things in the showcase hall.
As the conference started, I covered the keynotes for day 1 and day 2. The keynotes were interesting. With great talking points from Kelsey Hightower, Michelle Noorali, and many others. But of all the topics two high points stood out; Metaparticle and the fact Kubernetes 1.9 is boring.
Kubernetes 1.9, BORING!
The idea that the 1.9 release of Kubernetes is boring is a little misleading. It’s more accurate to say that Kubernetes is in a solid enough state, with few changes, and the 1.9 release is the first of a new trend of fewer breaking changes, increased reliability, and all those things that make software ready for the prime time. As if Kubernetes wasn’t already ready for the prime time eh! This however really does set the new precedent for Kubernetes releases — expect to see the majority of changes, progress, and the like in the tertiary libraries, plugins, and surrounding software of the Kubernetes ecosystem.
This project seems like a conversation I keep having with people. How to better deal with building systems, distributed systems specifically, and manage it all programmatically instead of oodles and oodles of configuration leaking out everywhere and breaching levels of concern. Yeah, Brendan has taken to coming up with a solution, by creating a general library solution around this. The solution he started he’s named metaparticle.
Within the last week between now and the conference I’ve found an article and code repo, and PR that I’ve found interesting about metaparticle. Here are a few I’ve collected:
Ryan Clair’s (Twitter) post on metaparticle is excellent, delving into the topic thoroughly and explaining the context in which metaparticle has been created. He also voiced some thoughts about my exact concerns I had, which is, is this the right abstraction layer? Is it providing a clear separation of concerns or bleeding them together at the wrong point of interoperability? Great post Ryan. I’ll be thinking a lot more on the matter too.
Gitter Chat which has an ongoing conversation going on about metaparticle.io. Check it out.
Overall I’m looking forward to the next KubeCon, and may even attempt to attend or speak at the Copenhagen KubeCon coming up. I’ll definitely be at the next KubeCon in the US, which should be easy since it’s in Seattle!
Day two kicked off (read on for day one wrap up) with Kelsey Hightower, Chen Goldberg, and Anthony Yeh. The big push from Kelsey and team focused their keynote around the development story around Kubernetes. Specifically, that a developers and apps users, should never need to know they’re using Kubernetes. He, Chen, and Anthony all talked about the idea we developers – as I’ll offer is true – want to work within our workflow committing, tagging, and knowing our applications will appear in test, development, QA, UAT, and production as we work.
I hope you’re having a good morning so far. KubeCon has kicked off in full force like the pro-conference that it is. With 4k+ people in attendance the crowds are distinctive, even among a city like Austin. The conference lit off the day with an absurdly early registration time of 7am, and a continental breakfast of some fruit and pastries.
The keynotes this morning kicked off at 9am. Dan Kohn, the executive director of the Cloud Native Computing Foundation (CNCF), started off the keynote session. He started off with a simple story about communities building. Outlining a quote from Tim Hockin, “existing time for boring infrastructure”. This really wraps up so much of everything happening in this space these days.
“Exciting time for boring infrastructure.”
Dan continued, pointing out that Linux is one of the largest projects on Github and probably one of the, if not the most important project on Github. Kubernetes is in the top 10 on Github too, as Jim Zemlin said, “Kubernetes is the Linux of the Cloud”. The commit and member involvement in Kubernetes. The growth of the KubeCon event and how this event in Austin is 4x the last 4 KubeCon events! Huge growth.
“Kubernetes is the Linux of the Cloud” – Jim Zemlin
Dan then introduced an unexpected speaker from Alibaba that elaborated on the massive scale of Alibaba, its leading position in China, and other projects Alibaba is doing with Kubernetes.
NOTE: Diveristy Scholarship
The US failed to provide visas for 4 of the diversity scholarship recipients. However they’ve been offered to attend the Copenhagen event, which in the end the US has taken more more hit because of our poor immmigration and border control rules in the United States. Something that needs to be modernized, and I’m sure we’ll keep losing out for years to come until national leadership has the spine to fix this. But I digress, onward with the keynote.
Michelle Noorali, Senior Software Engineer at Microsoft, then was introduced and came out to provide information on Special Interest Groups (SIGs), the deep dive sessions, techincal salons, and hallway tracks, and more details.
She continued to give us the story of KubeCon, Kubernetes, and the distinct history of how the rise of microservices, cloud, and container technology has changed the landscape of infrastructure and related technology. This is something important as the story isn’t always clear, but the story is a fundamental detail that informs and provides a more clearly defined path to where we’re all going with this technology.
Next Michelle dove into some specifics about the Workloads API coming in Kubernetes 1.9. For one it’ll be stable. Another is Windows Server container support beta, which I guess is stable – I haven’t used it and am curious who is, I’d love to talk!
Michelle then introduced Tom Wilkie to talk about observability and Prometheus.
Next up she introduced Eduardo Silva to discuss fluentd and announce the v1.0 release. Highlights include multi-process workers, sub-second time resolution, native TLS/SSL, otimized protocol buffers, and improved Fluentd protocol buffers. He continued and elaborated on the data streaming options, and flowing to technology like Kafka. Also, fluentd has been ported to Windows finally so Windows Server users can now natively use fluentd.
Reliability? What’s that? Oliver Gould then introduced by Michelle to talk about linkerd and give us some insight to the progress of the tool. Oliver then introduced Conduit too, and then gave us a demo of it working.
Michelle then came back around to wrap up her section of the keynote and cover some additional projects within the CNCF; grpc, envoy, and such.
Diane Marsh of Netflix came out next to discuss the importance of culture of building these tools, and what happens with these tools. The core of the talk focused around tools Spinnaker, Asgard – namely continuous delivery – and other Netflix OSS and how culture plays a part. She detailed how Netflix culture affected how people accepted and were able to use Asgard or Spinnaker. For instance, the culture at Netflix is one of freedom, and many companies don’t have this level of maturity. Netflix has this level of freedom, to deploy, because of a very strong level of trust.
Finally to wrap up the morning keynote, Adrian Cockroft came out to show us a few things. Leading of with cloud native principles; immutable code deployments, high utilization with aggressive efficiency by turning things off, pay as you go, no waiting globally deployed and distributed models. At AWS he’s working to increase the contributions AWS makes, working with CNCF (which they’ve joined), pushing cloud native, and integrating CNCF components (all those projects) into AWS.
Adrian also noted, importantly, that all AWS work with Kubernetes is upstreamed to Kubernetes itself and not a fork of the project. They’re working toward making all integration at all levels within AWS. Some of the work even include work and partnerships with Heptio around authentication for IAM within AWS. Lots of good things, and a lot of high integrity work!
Beyond Kubernetes and other elements Adrian’s teams are working on integrations with SPIFFE, HashiCorp Vault, and other open source tooling. I for one am pretty excited aobut these tools coming online at AWS as it’ll make my life easier to get some great things deployed and enabling customers and groups I work with! “Much excite, much wow” as the internet would say.
Adrian then dove into Fargate, discussed how it folds in with EKS, and how the integration is going to work.
I’m on another plane departing Seattle via SEATAC (SEA). An Alaska Air Boeing 737-900 to be specific. The flight is currently en route to Austin, Texas and the vast majority of people aboard are going to KubeCon. The seats, as they always are, aren’t built for any mortal, normal, reasonably sized human being. So we’re all cuddled up annoyingly but making the best of it we can. Seriously though, I’d rather be on an overnight train. I’d rather spend another 24+ or more hours comfortably studying some Netflix infrastructer and chilling out instead of flying, but that isn’t really an option in this giant country, so onward I go as the dream of comfort in transportation eludes me.
I’m setup and am aiming to provide coverage of numerous events, topics, and the like while at the conference. To boot, after conference I’ll be writing up some coverage of open source projects and companies that are at KubeCon.
There are a few new practices, techniques, and related things I’m trying out so I can cover even more of the event with useful things. We’ll see how that works. As always, much of my coverage will be on the various mediums I post to. The rest may appear in other various sources, which I’ll tweet and provide a summary email via my Thrashing Code News at the end of the conference.