Karimah & Aneesa | 15 November 2021
This is part 2 of our first KubeCon, to see the first part click here.
My name is Karimah, another apprentice at LiveWyer, and I am going to go through some talks that I watched from my first time at KubeCon. Similarly to Lorenzo I just completed my GCSEs and am really looking forward to learning more about emerging technologies, such as Kubernetes. We decided to attend our first KubeCon to get an introduction to the Cloud Native environment and get some context into why Kubernetes is useful and needed today, whilst working towards our apprenticeship.
This talk is about speaker Alice Wasko transitioning from a college student to a Support engineer working with Kubernetes. She walked us through the tips and tricks as well as skills on how she became a support engineer from a college student. She started with little to no experience and was put into a bootcamp to experience work with Kubernetes. Alice’s advice to us students was anyone with enough motivation and interest is enough to become confident in working with Kubernetes. She grew her skills through steps of knowing background knowledge, learning core skills then slowly introducing advanced skills.
I felt as though I could relate to what Alice was saying as I myself am a student and newly introduced to Kubernetes. I still feel like I am currently at the stage where I’m learning about the background knowledge, however I can definitely say KubeCon allowed me to experience an environment where there are students like me who are yet to venture into the world of Kubernetes.
Edge computing is a way to bring computation and data storage closer to the sources of data. Typically, they’re in sites that aren’t datacentres. Examples of locations include:
Jeff Sphar goes on to explain how Raspberry Pi edge computing works using K3s. Some of the benefits of using a Raspberry Pi include it’s small form factor, low power consumption and it’s ability to bootstrap applications with an active internet connection.
The talk was nicely tied together with defined project goals making it clear that the objective was to get to a K8s API as fast as possible whilst also making it highly available, inexpensive and with reproducible code.
This talk is the journey of the very well known music service provider Spotify, presented by the engineers Johan Haals and Patrik Oldsberg. Together they both explained how Spotify started off by using spreadsheets to track the state of their software and infrastructure, and how they created an internal developer portal called Backstage which supports machine learning models documentation, data pipelines and more.
Surprisingly, Spotify started off with only 6 people running all of their Operations. They spoke about adding more levels of automation, so that things that would take weeks or months could now be done in a single click of a button. Simply put, Backstage is a framework for building developer portals along with an ecosystem of plugins providing a set of common APIs. Backstage’s introduction of text documentation made it easier to write and find documents. Another one of the changes in Backstage was that they no longer handled just services, a whole different set of tools was now available.
As a user of Spotify I was very much interested to know how the service functions. I enjoy the service of podcasts it provides and knowing that Spotify really started off with only 6 people, managing all of their operations gives me so much hope that one day if I were to build a software, it will always start small.
The speaker uses an analogy of a bumblebee to explain GitOps practices to beginners. The analogy runs through ‘Granny Bee’ who owns a pastry shop and she asks her grandson ‘Git Bee’ to help her at the shop.
She tasks him in making three different flavours of cupcakes which she asks him to do everyday for the rest of the week. Git Bee decides to plan out his work for the week and he makes an ingredients list for each of the three different flavour of batter. Git Bee realises that in order for the other worker bees at the pastry shop to use his recipe, he would need to give them access to it. He decided to tag each of the raw materials so that his fellow worker bees could work on the same recipe without mixing the ingredients. This increased the productivity of the worker bees.
This analogy was useful in understanding the core GitOps practices. For example, defining the resources that are required, tagging resources and being able to collaborate simultaneously on the same project.
In this talk, Liz Rice introduces us to Extended Berkeley Packet Filter (eBPF) showing us the powers it has in making the kernel dynamically programmable. eBPF programmes are attached to an event which triggers the programme. These events could be anything from the arrival of a network packet to a user application making a system call.
The talk followed a simple hello world style demo showcasing how eBPF can be used for tracing by attaching a ‘Hello KubeCon’ message to processes running across a machine. This really highlighted that eBPF was made aware of everything happening across the kernel. To ease our mind about potential security concerns, the verifier component was also introduced to us which prevents issues such as infinite loops from occurring or memory being exceeded.
It was exciting to see the power of eBPF through a few CNCF eBPF projects; Falco, Pixie and Cilium. These outlined that eBPF can be used to provide new approaches in the areas of networking, security and observability.
I really enjoyed my time at KubeCon and the variety of talks has made me more excited to delve into the various areas of Kubernetes. I liked that there was a mixture of technical, operational and community based talks. I can’t wait to practically experience using Kubernetes myself and to hopefully enjoy more advanced talks at next years KubeCon.