Kubernetes Community Day 2019 Amsterdam

The Kubernetes community day 2019 was the second time around for this annual event, which started last year. This time around featured speakers where next to the local Dutch users, also a few of the international speakers, including Dan Kohn (CNCF), Diane Mueller (RedHat), Ralph Squillace (Microsoft) and many more.

Topics

Evolution and historic overview, User (war-)stories, How-to's and Compliance.

opening

Looking back on how a good idea is still difficult to get of the ground, thanking people of the first hours, Dan Kohn also picks up on community, stating that this is the first worldwide Kubernetes community day (CNCF sponsored that is, I'm sure there are loads of community days around Kubernetes already).

community

Both Diane and Dan stress the idea of community, while Diane does this literally, Dan does this by saying communities are build around ideas "in the air". Standing on the shoulders of the greats before you is a recurring theme, at least before the break..

evolution

The idea of evolution is a recurring theme as well, based on community building, getting to know the people and the projects, in this Cambrium explosion of ideas. Everybody is approaching needs and solutions in a different complementing, sometimes repelling or contradictory way. That's one of the reasons this community thrives (IMO), but also something which has to be well nurtured, also at this community day. Don't let the tools fool you, it's a community that makes the movement.

evolution use cases  

The next speaker also gives us a nice insight that also a consumer of Kubernetes and/or Cloud Native technology goes through an evolution of sorts. Especially usefull is the diagram below where you can depict solutions on a DIY <--> Managed horizontal axis and a multi-cloud/cluster and single cluster vertical axis. Looks like a helpful way to plot solutions and see how solutions from one vendor to another differ from DIY to payed and managed. Organisations can also see where they'd want to sit in this diagram, asses their apatite for low level control vs full service.  

Kinvolk.io helped a customer evolve from a vm-user on AWS to a bare-metal Kubernetes user. They were hurt in their business case, because of costs of vm based infra costs, these costs did not scale well with a growth of customers. Especially the egress network traffic was a hurting them a lot. Shifting to supported Kubernetes on bare metal and extending this back into non-egress heavy application parts (which remained in AWS), was the solution. The evolution of a managed cloud to a DIY/Packaged/Supported roll-ur-own build had to tick a lot of boxes, but it is possible and like the "k8s the hard way", will give you flexibility but also has possible support-ability options.

user case AH

Albert Heijn showing how they have setup their environment and how difficult it was to let go of the pets as an organization. Solid overview, nothing to fancy, but with an emphasis of transforming from a pet-shop to a cattle herder, less petting and more hands-off.  

operator (framework) evolution

Operators are used as an extension pattern to help run new workloads in Kubernetes without adjusting upstream Kubernetes API to a specific use case. But how do you create, adjust and manage your operators? How do you make sure when one operator changes that this does not effect other operators? This also touches upon the multiple ways you could deploy management functions. Although the choice to even evaluate operators is a decision on and of itself, there are more Kubernetes patterns which could be used, especially if you don't need to access this management functionality through the Kubernetes API. This might be because it's a session about the Red Hat tooling, which they developed to manage operators and as such this question is of course something you would have passed already when looking for a solution to manage your operators...

compliance as code at ABN AMRO

At ABN AMRO they're trying to leverage the way code gets deployed and integrate it in a way that compliance is automatically maintained. In a lot of companies speed of deployment is dragged to a halt (sometimes even going in reverse) to the forces of regulatory compliance. ABN AMRO tries to get ahead of that curve by operationalisation of compliance into deploy-able code, which becomes an integral part of their deployment pipeline. Well rounded presentation, with nice insights into their road map and what works for them. This is also how I would like to implement architecture, by making it difficult to sway and just not getting past the auto-tests and deployment checks. Halting the package if and when it does not pass tests based upon agreed patterns and principles.

lightning talks, or showcases..

  1. Missed the Shipper lightning talk,
  2. Mambu ArgoCD - still have to get it running myself, liked a few features, also remember the tool from a TGIK (Heptio) session by Joe Beda,
  3. Climate action lightning talk connects to the vegie barbecue, I liked the elaborate vegan options, as someone not eating meat and consuming dairy products. Also a plug for (of course) a technological call to action https://climateaction.tech.
  4. hackyourfuture.net - how welcoming are you for new comers, who just learned the theory of coding, but need practical work experience, the need to bring this into our professional environments, our companies.

cloud, the edge of..

What's a cloud? According to Microsofts Ralph Squillace a cloud exists by the grace of it's edges and Kubernetes is the ground floor of what everybody will build upon. It should get easier for you to provide your organization with that floor to build great things upon. The examples of the mentioned software proofs that you can really run Kubernetes anywhere on any device:

  • K3s on Pi (and/or laptop)
  • KIND on your laptop in a container (creating a multi-node cluster on your laptop, in containers, running containers.. (inception much..?))

The discussion was entertaining, especially to the members in the audience nearing 50 years of age and above. A lot gets overlooked if you only focus on the-new-thing and not paying attention to what we as an industry already learned.

Azure-stack deployment using Terraform

Not the promised Terraform talk from the title, but did gain some new knowledge on tactical Azure-stack (by Dell) disconnected from Azure AD, so it might be worth a look, although it's always a journey with MS products to see which parts are actually production ready and which are still at the pre-alpha customer testing stage. Not that many success stories about Azure-stack, especially the Azure AD tie-in remains a deal breaker..

looking forward and back

Pieter Lange throws in his psychic skills and promises bold claims about the future! Doing things well is still something which needs a lot of care and attention, just because you use new tech, it doesn't invalidate old best practices from the 25 year Linux, distributed systems management experience. Use Kubernetes policies and RBAC, it might just save your company... and of course Open Source won, even though I won't claim the desktop (as most desktop OS-es are pieces of shitty functionality on hardware which is already obsolete while you're unpacking it and what do you need a desktop for if you can run cloud native anywhere and anytime..)

great event, at a great price-point

Although we do talk a lot about community building, a lot of focus remains on the fact that Kubernetes, Cloud Native and the eco-system is also just business. That said it's fair to mention that I always prefer OSS community events to the grey-men-in-blue-suites events held at so called "vendor neutral" gatherings. Although Microsoft and Red Hat seem to be the dominant featured vendors at the event (and AWS is, as always, notably absent...), it does not come across as such, nice balance of showing what you do and that you're part of a community.