Well, I think this blogpost will start similarly to the one I wrote about my first time attending DevCon back in Mauritius. "Writing this, I'm both really happy it happened and really sad it's over. I guess that's how everyone feels at their first DevCon."
If you noticed that I haven't been writing or tweeting much these past few weeks, it's because I've been planning for this behemoth of a conference. Well, I'm back at the AirBNB and the conference is over, so I feel like writing about my experience before the only thing I remember is the massive SUSE chameleon.
If you didn't know already, I've been playing around with Kubernetes on my homelab for the past 2 or so years. I am a fan of K8S and the whole ecosystem of apps built on top. And for me, KubeCon (& CloudNativeCon), was the most amazing experience imaginable. In fact, this blogpost doesn’t do justice to the scale and awesomeness of the event.
Well, that's three paragraphs of rambling, let's move on to KubeCon itself!
Day 0
Yes! There's a Day 0. Well, technically speaking, there's also a Day -1 but that was only for picking up your badges, so I won't go into detail on the uninteresting story of landing in Paris, and dragging my airsick backside to the conference center to pick up my badge.
Day 0 was mainly made up of CNCF-hosted co-located talks, Sponsor-hosted co-located talks, as well as a flurry of Lightning Talks.
The CNCF-hosted co-located talks (god, that's a mouthful) required the fancy all-access pass and you could realistically only attend one since they all happened in parallel for half the day. These were events targeted towards specific technologies or concepts, such as Argo, Cilium, Istio, and observability, as well as many others.
I decided to attend Rancher Day as it was only $20 extra during registration, and I use Rancher tech for my homelab in the form of K3S, the lightweight K8S distribution. Additionally, Rancher got procured by SUSE, which gave me an extra reason to believe the session would be good.
Rancher day was an awesome workshop event and I could not have hoped for a better start to my first-ever KubeCon. The first part of the day saw Erin Quill take the stage to show off Rancher Prime, Rancher’s GUI for cluster management, with easy application of Helm Charts in barely a few clicks. Despite keeping things very simple, Rancher Prime also allows lower level access to the cluster, letting the administrator deploy or troubleshoot apps through the usual kubectl commands.
After a very big lunch, we went back for Rodeo No. 2 covering NeuVector, a container security platform for 0-trust regarding Kubernetes. It is a whole suite of tools which can enhance security in a production cluster. One of them scans the processes/commands running within a container and then can block any commands running from the pod which don’t fit the sequence scanned. This would prevent someone from accessing a pod and escalating by querying the Kubernetes API from within the pod, or trying something bizarre with the database - behaviour not present in the container itself when it was scanned.
After Rancher Day, I made my way over to the main conference centre to attend the project lightning talks for the different projects in the CNCF landscape, followed by more lightning talks. I particularly resonated with Minecraft as a means of education, since the rabbit hole I entered when I researched how to run a Minecraft server for me and my friends is the very path which brought be all the way to Paris!
Day 1-3
On the first day of the core KubeCon event, I was up really early due to my body clock being 3h ahead. This gave me enough time to get ready, have some breakfast, and walk to the conference centre. I was going with my mum, who got one out of 50 free tickets from the CNCF for those affected by budget freezes or layoffs. Lucky too, as she is now knowledgeable enough in Kubernetes to grill the guys in charge of K8S at work.
All three days started with the keynote presentations. These were held in the massive presentation room, with screens setup all over the place to broadcast the slides and the speaker all the way to the back. From a seat at the front, it was nearly impossible to see the back! I felt like a mouse in the giant’s den, compared to the much smaller DevCon back at home. The keynotes were quite nice, covering a variety of topics including AI on Kubernetes, sustainability, and diversity in tech.
After the keynotes, I headed down to the solutions showcase which had just opened. This was a whole floor dedicated to the sponsors of KubeCon, as well as the project booths for different Cloud Native projects such as Longhorn, ETCD, Flux, Istio, and the KCD.
Most of my time was spent on this level, talking to different people from the CNCF projects, as well as representatives from companies at the vendor booth. I found these conversations extremely important to learn about new technologies or best practices. I also made a list of promising projects to try out in my homelab, be it logging & troubleshooting tools, Redis replacements, Web Assembly, and many more. In fact, once I get home, I’ll be getting very busy playing with these tools and really pushing my cluster to the next level. Fingers crossed that I have enough processing power on it!
I won’t go into much detail on the talks I attended, but here are some of my highlights.
- Rust - I attended two talks on Rust, both of which were very interesting. I am in the process of learning the language, and seeing what Rust can accomplish, being used in Linkerd, is fascinating. Without compromising performance for safety, Rust may very well be the future! Now is just the complicated task of learning the complexities of the language haha
- Longhorn - Longhorn is the block storage storage solution for Kubernetes, offering performant and redundant persistent storage. I attended this talk as I already know my way around Longhorn, using it on my cluster, and wanted to get a deeper understanding in its low-level workings.
- OPA - I discovered OPA and this talk thanks to Anders Eknert who is a mutual on Mastodon. It was the first time learning about OPA and I’m very glad to have attended. Kudos to Anders and his co-speaker who made the session such that it catered for newbies like me, and the more advanced users. OPA is a super straightforward and global way of authentication which can avoid each department of a company implementing their own version in their own language - which would complicate administration.
- ArgoCD - I attended the workshop hosted for ArgoCD on the last day. It was a very fun event which placed you in a sandbox and had you use ArgoCD to roll-out changes to a deployment in blue-green and canary mode. CD tools like Argo and Flux are really important tools to have in your cluster to simplify deployments and avoid a folder of manifest files.
- Web Assembly - Fast, compact, architecture agnostic? Web Assembly seems to be the future of containerisation. Being able to compile my app into a WASM binary which is ultra-light and then deploying it to Kubernetes may be the next step, promising lower resource utilisation and thus a greener means of serving your software.
For the projects and vendors I got to meet, here’s a short list of those that caught my eye. Of course, there's all the big names like AWS, GCP, Intel and the like. But we're all too familiar with them, so I sought out the smaller companies which are starting to make a name for themselves.
- Dragonfly - a source-open drop-in replacement to Redis. Amidst Redis pivoting towards the gen-AI hype train, Dragonfly may be an interesting alternative. The source code is over on GitHub and despite the limitations of being a drop-in replacement, Dragonfly achieves staggeringly better performance than Redis.
- Cosmonic/wasmCloud - Cosmonic is a cloud service provider for web assembly. I’m including them here because they are a pretty cool company and the guys there were super chill. However, you can always use wasmCloud on your own hardware with Kubernetes.
- Honeycomb - Built for observability, Honeycomb unifies logs, metrics and traces to provide a seamless interface to debug and optimise your entire cluster from a single UI. They offer a generous free tier of 20 million events per month, with no time limits.
- CircleCI - While Github is following the AI hype, CircleCI is doing what it’s done best for ages: CI/CD. They have a very generous free tier as well, with 6000 free build-minutes per month! They also claim to be on average much faster than their competitors.
- NodeShift - I’m not one to follow on ISO standards for cloud platforms but my brain was nonetheless buzzing after listening to the guys there talking about their compliance and such. They offer a cheaper alternative to the cloud monopolies (Google, Amazon & Microsoft) with nearly the same compliance standards. While they don’t have any managed Kubernetes clusters yet, you can always setup a bare-metal cluster on their cheap VMs.
- Kraftcloud - Fast and efficient is the name of their game. On their booth, you had a green button to spawn a new instance of a web server, so of course I obliged. After pressing the button, the new instance is created, runs for a second or two, then goes to sleep. Afterwards, I can try to request something from the server, at which point the web server comes back to life, responds to the request, then goes to sleep again. This is awesome as it’s really scaling up and down in the microseconds, leading to no noticeable delay for the client. I am not yet too sure about their self-hosted offerings but it seems you can run their tech on Kubernetes.
As well as meeting the vendors, I got to meet quite a few people over at the project booths.
- Over at Longhorn, I met David Ko from SUSE, who was more than happy to talk about this awesome piece of technology. I often say that Longhorn is the coolest piece of software running on my cluster, with reason!
- At the OPA booth, I got to meet with Anders Eknert, who I had interacted with on Mastodon prior to KubeCon - in fact it started by him suggesting his talk when I bemoaned the overwhelming number of sessions at KubeCon.
- At the KCD booth, I got to meet Julia Morgado who is a CNCF ambassador. Her YouTube channel is a must-visit for anyone attending KubeCon for the first time, and her vlogs helped prepare me for what to expect out of the conference. I also got some information on KCD's - Kubernetes Community Days - resembling the MSCC and FrontendMU monthly meetups back home. Both me and my mum also sought out information on setting up your own Cloud Native User Group. It would be awesome to have our own group like this in Mauritius to promote awesome tech like Kubernetes.
Additionally, if your UG is approved by the CNCF, then you have exclusive access to resources, as well as discounts for material online!
France
I figure I wanted to add a little bit about Paris/France itself, especially given how different life is there compared to Mauritius. First of all is the really extensive rail and bus system forming a web across Paris. Between trams, the metro, and the RER, it is extremely easy to get around over long distances for very cheap (2 Euros per trip) and whatever isn't covered by the rails is reachable through the buses. And despite covering such a large area with countless trains/buses, the whole system is extremely clean and the interval between rides is extremely little - 5 minutes at most.
Furthermore, Paris was so much more friendly to pedestrians and cyclists, and I would often joke with my mum about how cars seemed to be the least priviledged road users XD. And not only could I walk/cycle anywhere without fearing for my life (as in Mauritius) but everything I needed was reachable without needing a car. Admittedly, this may be an advantage of being towards the centre of Paris (within zones 1-3) but it was nonetheless incredible to have numerous restaurants (serving healthy food) and supermarkets within 5 minutes of walking! Actually, with all the walking I did, coupled with how healthy the fast food was, I wouldn't be surprised if I lost weight during my week in France!!
À La Prochaine!
And on that note, this is it for my blogpost on Kubernetes. I have a LOT of things to try out, as you can undoubtedly see for yourself. I’ll be quite busy when I get back to my cluster, between trialing new tech and pursuing the CKA exam. Speaking of, I was given the “Acing the CKA” book by the amazing Chad M. Crowell before the book signing.
And once I’m “done” with all of that, it will be time to get ready for next KubeCon in London 1-4 April 2025! Hoping to either win the lottery or sponsorship/scholarship by then :P
