Streamlining DevEx with K8s Dynamic Environments and Mirrord

In the talk we will show you how we at PerfectScale run dynamic environments for our microservices, spinning up required infra, secrets, and all other needed resources. We are using Helmfile with helm charts pulled directly from Git repos, inline kustomization, advanced templating, secrets encrypted with sops, while allowing easy filtering which releases to install, state initialization, and other features.

All this is made available from CLI or GitHub Actions. But this is just half of the way to improving developers' experience. We don't want to spend expensive minutes just waiting for full git->ci->cd flow while developing the application. Instead, we are using MirrorD for a fast feedback loop. Together with dynamic environments, we can provide fully featured env which is equal to prod (except customer data) with the ability to work with it as with local and run/debug your local application enjoying in-cluster services access. We will share our experience with Telepresence and MirrorD, explain why we chose it, and how we overcame issues. We would also like to present our new open-source project which helps to improve PR reviews from Product and Design teams with Preview Environments.

Tools which we will talk about:

- Helmfile, Helm, Kustomize
- Sops
- MirrorD
- Preview app comments
- ArgoCD
- Makefile
- GitHub Actions

Join us to learn how to enhance your development workflows with Kubernetes and these powerful tools, and see firsthand how we streamline dynamic environment management and rapid feedback loops at PerfectScale.

The A to Z of Kubernetes Performance Optimization

In the talk I will provide a tight and practical full lifecycle overview of Kubernetes performance optimization:
- optimized images
- optimized nodes
- optimized resource allocation techniques
- optimized binpacking
- alternative runtimes
- reliability concerns
- the cost of optimal performance
- how to measure and iterate
No vendor pitches, no fluff - just practical battle tested advice from the trenches.
Let's get your clusters optimized!

The K8s Operational Excellence Playbook

How proud are you of your Kubernetes clusters? Are they ticking like well-oiled clocks or is entropy slowly but surely wearing them down and you find yourself constantly firefighting just to keep them operational. Or are you - like most of us - somewhere in the middle ground - just doing the necessary minimum so you can get a good night sleep?
It doesn't have to be that grim. Let me layout the iterative, step-by-step process to get you to Kubernetes Operational Excellence.

Based on more than 5 years of implementing and maintaining K8s clusters in Production.

Platform Engineering with Backstage

Backstage is a CNCF project.
It's the leading open-source platform for building Internal Developer Portals - systems at the core of modern platform engineering.
In this workshop we'll learn the basics of the Backstage platform and see how to customize it for your software organization's needs.
Some topics we will cover:
- Service scaffolding
- Knowledge sharing and documentation
- Service Catalog
- Kubernetes integration

K8s load testing at scale with k6-operator

Grafana k6 is an open-source load testing tool that makes performance testing easy and productive for engineering teams. k6-operator allows running load tests in a distributed manner with additional perks such as using xk6-disruptor for fault injection (chaos engineering style)
In this talk we'll present what we've learned using k6 to load test our SaaS application workloads.

A Platform Left Out in the Cold

The platform buzz is at its strongest right now. Many companies invest time and money in building internal delivery platforms only to find themselves struggling to win developers adoption.

This talk will present the main psychological and cultural hurdles standing in the way of platform adoption and the ways to tackle and prevent them in the long run.

Best practices in designing, planning and budgeting internal platform initiatives will be laid out with practical examples and stories from the field.

FinOps for K8S : Auto-scaling Kubernetes on a Budget

One of the most exciting features of Kubernetes is its ability to scale up, down, out and in - both on demand and automatically. But scaling is first and foremost an economical decision.
Now that so many organisations depend on Kubernetes in their production environments - it's time to define the best practices of how, why and when to scale. And analyze the involved costs - on the operational, cognitive and environmental levels.

We'll go over the Kubernetes scaling technologies and solutions, understand the economics of a Kubernetes cluster and discuss industry proven practices for Kubernetes performance, cost and sustainability optimization.

Manage Cloud Costs to Make The World a Better Place

Reducing cloud spend can also be great for ecology and sustainability. But there's much more to this than just managing your cloud bill size. Let's dive into the more intricate details of the complex interrelationships between cost, resource consumption and your data center locations.

The Promise of WASM

WebAssembly (WASM) is a super-promising technology that enables secure and efficient distribution of binary extensions. WASM binaries are especially great for serverless, edge and IoT deployments due to small size and high performance. In the session I'll provide an overview of WASM, discuss its practical applications and its potential of changing how we deliver software.

Building Platforms That Scale

Platform engineering has become the de-facto standard implementation of the DevOps approach. Internal platforms are becoming the backbone of effective software delivery. But the discipline is still very young. The principles of platform design are still evolving.
One basic requirement is that an internal platform stays flexible and scalable - an enabling entity rather than a hurdle.
Let's review the industry's lessons on how to build platforms that scale.

Some pods are bigger than others

Modern Kubernetes workloads are highly diverse. From always-on lightweight web services to gpu-intensive ETL workflows. Managing resource allocation for all these is increasingly challenging. Let's dive into the world of auto-scaling, pod QoS and priority classes, hpa, vpa and more - ensuring all pods can live together happily in our clusters.

Does size matter? The Joys and Pains of K8S (Auto)Scaling

Autoscaling is one of K8S "magic" features that folks get especially excited about. But beyond he technical details - autoscaling is a business concern. In the end it all boils down to balancing efficiency with availability. How much resources do our containers really need? When do we add more pods? How many pods is too many? Do node sizes matter? Let's see how to scale Kubernetes the scientific way.

Stress and Burnout in IT - Stories we Don't Tell

DevOps is supposed to make our life better. And it frequently does. But there are too many casualties to keep silent. I've burnt out myself in the past and until this day I see folks stressing out, getting depressed and demotivated. Even when they "do DevOps" on paper.
This talk will present burnout stories in order to better understand how "DevOps" can be stressful and how to prevent this from happening,

Resilience in Engineering... and Life

Resilience starts with acknowledging failures. The systems we're building and the distributed, global organizations we're part of are highly complex. Failure in complex systems is a given. While resilience defines our persistence and ability to use failure for growth and learning.
Let's discuss how to build that capability on the technical, organizational and finally - personal level.

Platform Engineering vs. SRE vs. DevOps - do we have to choose?

A flurry of clickbait posts enflamed the internets during the last year proclaiming "DevOps is Dead, Long Live Platform Engineering"
Like many of us - my original reaction was to disregard this as another marketing fluff.
But - everything happens for a reason.
Let's hold a panel to discuss why this happened, how DevOps is changing and what we can expect going forward.

Build a Platform Not a Barricade

Platform Engineering is all the buzz now. Centralized developer platforms are touted as the panacea for complexity, security breaches and lack of efficiency. But centralization often brings bureaucracy and loss of autonomy. How to avoid our platform team becoming yet another silo surrounded with walls that DevOps was supposed to bring down?
In the session I'll outline the pitfalls and recommendations for building internal platforms. All based on my 23 years of industry experience.

The art of Kubernetes resource management and auto-scaling.

Modern Kubernetes workloads are highly diverse. From always-on lightweight web services to gpu-intensive ETL workflows. Managing resource allocation for all these is increasingly challenging. Let's dive into the world of auto-scaling, pod QoS and priority classes, hpa, vpa and more - ensuring all pods can live together happily in our clusters.

A hands-on workshop

Kubernetes Performance Tuning Workshop

Kubernetes resource allocation and performance tuning is not a walk in the park.
In this workshop we'll discuss and get hands on with:
- Allocating resources to our workloads
- Defining sane LimitRanges and ResourceQuotas
- Understanding VPA recommendations
- Balancing resource cost with resilience
- Defining scaling factors
- Looking at additional recommendation and performance tuning techniques.

Practical Kubernetes knowledge is a pre-requirement.

A full-day or half-day workshop

The Intricacies of Container Economics

Most container clusters out there aren't optimal in either their resource utilization or performance. Waste and technical debt are accumulating.
Optimization starts with understanding the basics of container resource management. Let's dive into application performance, resource allocation and how to measure and test these correctly.

Topics covered: cgroups, cpu, memory, iops, orchestration, kubernetes

Non-Violent Communication for Better Software - a Workshop

Engineers who are happy build better software. Great teams thrive on great communication.
I've led tech teams and coached numerous software organizations to streamline software delivery. Every time I found that ineffective and straight up violent communication is one of the main blockers to productive collaboration.
Today I'm also a certified non-violent communication (NVC) coach.
In this workshop the participants will learn empathic communication techniques for better engineering experience and collaboration and participate in exercises to practice applying these techniques in real life.