TNS
VOXPOP
Do You Resent AI?
If you’re a developer, do you resent generative AI’s ability to write code?
Yes, because I spent a lot of time learning how to code.
0%
Yes, because I fear that employers will replace me and/or my peers with it.
0%
Yes, because too much investment is going to AI at the expense of other needs.
0%
No, because it makes too many programming mistakes.
0%
No, because it can’t replace what I do.
0%
No, because it is a tool that will help me be more productive.
0%
No, I am a highly evolved being and resent nothing.
0%
I don’t think much about AI.
0%
Kubernetes / Observability / Platform Engineering

Platform Engineering Rules the Day: Eight Key Themes

With platform engineering at the helm, the future of cloud native development is poised for unprecedented growth and transformation.
May 16th, 2024 10:30am by
Featued image for: Platform Engineering Rules the Day: Eight Key Themes
Feature image via Unsplash.

Platform engineering has definitely taken flight. Indeed, platform engineering was the big topic at KubeCon + CloudNativeCon Europe 2024 in Paris a couple of months ago.

There were 60 sessions related to platform engineering, almost twice as many as sessions on large language models (LLM). Unsurprisingly, we find “developers” all the way at the top of the 20 most important topics at the event, with data and security following close behind. This makes sense as almost any session on platform engineering looks at how to enable developers to securely create increasingly intelligent data-driven apps.

Next come operations, integration and compute as topics that are also crucial in the context of platform engineering, reflecting the comprehensive approach required to support the entire software development lifecycle. These topics underscore the importance of operational efficiency, seamless integration of tools and services, and the effective management of computing resources.

Together, these topics highlight the multifaceted nature of platform engineering, aiming to streamline development processes, enhance security measures, and leverage data effectively, all while ensuring that operations run smoothly. This focus on developers, data, security, operations, integration, and compute demonstrates the holistic view that platform engineering takes toward improving developer productivity and facilitating the creation of sophisticated, data-driven applications in a secure and efficient manner.

Platform Engineering at KubeCon: 60 Sessions, 20 Topics and 8 Key Themes

The 60 platform engineering sessions at KubeCon Europe followed eight key themes, all focused on optimizing different aspects of the software development lifecycle.

Enhancing Developer Experience through Internal Developer Platforms (IDPs)

A pivotal theme in platform engineering is the concerted effort to improve developer experience via the implementation and optimization of Internal Developer Platforms (IDPs). Building multitenant, cloud-agnostic platforms with an emphasis on striking a balance between development velocity and system reliability is key. Topics at KubeCon covered a wide spectrum of real-world customer stories ranging from embedding security and observability tools, to demonstrating pragmatic approaches to debugging within the Kubernetes ecosystem. The integration of such platforms amid corporate mergers, underlining the cultural and technical challenges of aligning disparate engineering teams within a unified IDP.

These discussions hone in on the theme of cultivating internal platforms that aim to abstract away infrastructure complexity, enforce best practices, and allow developers to concentrate on delivering high-quality software efficiently. Additionally, generative AI has the potential to revolutionize platform engineering by creating customizable abstractions, further easing the burden on developers.

Interesting Products

GitPod Open Source or Cloud Hosted

I liked seeing GitPod as a sponsor of Cloud Native Computing Foundation (CNCF) platform engineering day as I have followed this specific product since KubeCon 2022 in Detroit and absolutely love how their cloud development environments provide a desktop-like developer experience, without clogging up an actual developer machine. I ran a couple of my Python-based Streamlit data apps on GitPod and found that no network configuration is needed. I simply run my app and GitPod hands me back its URL.

The fact that I can use my favorite IDE, Pycharm, and that GitPod installs all the needed dependencies for most apps without developers having to worry about it, are two additional plus points. Platform engineers receive central config and secrets management, standardized reliable development environments, and built-in single sign-on with popular platforms like GitHub, GitLab, MongoDB, AWS and Visual Studio Code.

Octopus Deploy

Octopus Deploy emphasizes ensuring uniform deployment across various environments — be it hardware, virtual machines (VMs), container clusters, or serverless contexts — spanning data centers, cloud infrastructures, and edge locations. It leverages the pipeline-as-code approach to streamline and maintain consistency in deployments, while also providing platform engineers with comprehensive visibility and control over the entire deployment process within an organization. The integration of CodeFresh, with its team of ArgoCD open source project maintainers, into Octopus Deploy further enriches the platform. This integration allows for a cohesive deployment strategy for Kubernetes applications, facilitated through a managed Argo platform, thus enhancing the platform’s capabilities in orchestrating and managing Kubernetes deployments efficiently.

Data Management and Stateful Applications in Kubernetes

As platform engineering matures, a notable theme is the provisioning and management of data-centric workloads within Kubernetes. Presentations and workshops at KubeCon Europe focused on the unique challenges and innovations in handling stateful applications, highlighting strategies for meaningful compute and storage separation, utilizing various storage technologies, and building resilient platforms. Kubernetes’s growing capabilities are emphasized, addressing topics like NVMe SSD limitations, automated volume resizing, and handling diverse analytic databases at scale. The theme delves into the practicality of leveraging Kubernetes’ native storage features to craft dependable and scalable data platforms, enabling organizations to utilize the full potential of their data in a cloud native environment.

Interesting Product

Vitess and PlanetScale

PlanetScale distinguishes itself as a distributed, cloud native database platform, built upon the robust foundations of Vitess and MySQL. The platform is particularly compelling due to its focus on a developer-first approach, facilitating the creation of applications for globally distributed SQL databases with seamless, zero-downtime migration capabilities. This is coupled with a straightforward and adaptable API, reminiscent of the ease provided by NoSQL platforms such as MongoDB. Vitess powers PlanetScale’s ability to execute database deployments or migrations — complete with rollback options — without incurring downtime or data loss, ensuring continuous operation and data integrity.

Machine Learning and Generative AI in Kubernetes

The rapid growth of machine learning and generative AI workloads presents unique scalability and operational challenges, especially when deploying large models with extensive computational demands. This theme reveals how Kubernetes, along with frameworks like Ray, facilitates the efficient serving of these AI models through seamless integration with hardware accelerators. Insight into the enhancements and cost-effective strategies for deploying generative AI models in a Kubernetes environment is provided, offering practitioners a path to infuse more intelligent and automated capabilities into their platforms.

Interesting Product

Ray by Anyscale

Ray is a distributed computing framework designed to scale machine learning (ML) and artificial intelligence (AI) workloads efficiently across clusters, offering a unified solution for a wide range of applications from deep learning to reinforcement learning, hyperparameter tuning, and model serving. Its core value proposition lies in its ability to effortlessly scale complex workloads with minimal effort, thanks to its simple, flexible API that integrates seamlessly with popular ML libraries like PyTorch and TensorFlow. Ray achieves this by allowing developers to specify computational resource requirements at a granular level, such as CPU and GPU allocations for individual tasks or actors within a job, facilitating optimal resource utilization and performance. Trusted by leading AI teams and organizations, Ray addresses the scalability, flexibility, and efficiency challenges in deploying large-scale, computationally intensive AI models, making it a cornerstone technology for modern AI-driven applications.

There is a commercial version of Ray offered by Anyscale. Anyscale provides a managed cloud service for Ray, which is designed for organizations that prefer the convenience and speed of a managed service over self-managing their own clusters and infrastructure. This commercial offering includes additional features, support, and services beyond what the open-source version of Ray provides, catering to enterprise needs for productionizing and scaling AI and Python workloads. Anyscale, as the lead commercial backer of Ray, aims to simplify the process of creating, running, and managing Ray workloads, making it easier for companies to succeed with AI and get value from their AI initiatives.

Advancements in Platform Abstraction and AI-Driven Development

In the swiftly expanding landscape of cloud native technologies, platform engineering is progressively centering on creating proficient abstractions and implementing AI-driven development methods. KubeCon Europe sessions on varying levels of platform abstraction delineate the importance of balancing education, feature development, and speedy implementation. The evolution of these abstractions is a testament to the industry’s ongoing efforts to refine the developer experience.

Concurrently, the use of generative AI to elevate platform engineering — via automating the creation of higher-level abstractions and APIs — further reinforces the trend towards easing the complexities involved in building and consuming these technologies. Leveraging LLMs to generate code and crafting more intuitive, human-centric APIs places platform engineering on a trajectory that promises to substantially diminish the barrier to entry, increase efficiency, and customize solutions suitable to organizational needs.

Interesting Products

GitHub Copilot and GPT-Engineer

GitHub Copilot is so effective at enhancing developer productivity as it allows OpenAI’s GPT LLM to “see” the project context of all project code. Copilot integrates well with Visual Studio Code and with Pycharm, my personally preferred IDE. Users will notice the AI continuously improves within an IDE project workspace the more it is used and the more it can observe how users adopt its code recommendations.

In addition to Copilot, you should look at GPT-Engineer. GPT-Engineer is based on GPT-4, but it is able to create complete, well-structured, apps, based on a simple prompt file maintained by the user within the project directory.

Kratix by Syntasso

Kratix is essentially a construction kit for Internal Developer Platforms. It is an open source framework to empower platform engineers with the capability to deliver Platform as a Service (PaaS) solutions. Kratix focuses on enabling platform engineers to build better platforms by harnessing the power of Kubernetes and other cloud native technologies. It allows for the creation of “Promises,” which are configurations that service specific requests, such as environment creation, within a Kubernetes cluster. These Promises can automate and streamline various non-development activities, making it easier for developers to request and receive the resources they need without direct intervention from the platform team. This approach not only saves time for platform engineers by reducing the need to service individual requests manually but also enhances the developer experience by providing a more efficient and self-service workflow. Syntasso offers an enterprise version of Kratix, with enterprise support included and consulting services available.

Cloud Native Networking and Security with eBPF and Cilium

The evolution of cloud native networking and security is a central theme, with a deep dive into eBPF (Extended Berkeley Packet Filter) and Cilium, a networking, observability, and security solution for Kubernetes. Numerous KubeCon Europe sessions explored how eBPF and Cilium are transforming the way containers and microservices communicate and are secured, spotlighting their scalability, performance, and flexibility. Use cases of production deployments, technical deep dives, and discussions on advanced features like API gateway security and encryption demonstrate the comprehensive capabilities eBPF and Cilium offer. Opportunities for learning from expert users and maintainers are also highlighted, emphasizing the community’s role in knowledge sharing and innovation within this domain.

Interesting Products

Cilium Enterprise by Isovalent (acquired by Cisco)

While Cilium itself is an open source project, Isovalent offers an enterprise version of Cilium that includes additional features, support, and services for organizations deploying Cilium in production environments. Cilium Enterprise adds enterprise-focused capabilities to the core eBPF-powered networking, observability, and security features found in the open source version of Cilium.

Falco

Developed by Sysdig and now a part of the CNCF, Falco is a cloud native runtime security project. It uses eBPF (and optionally kernel modules) to monitor system calls in the Linux kernel to detect anomalous activity and alert on threats at runtime. Falco can be used to secure Kubernetes clusters as well as containers and cloud native applications.

Multitenancy and Data Platform Scalability on Kubernetes

Ensuring scalability and resource efficiency when running multiple tenants and extensive data applications on Kubernetes forms the crux of this theme. Sessions delve into practical experiences and innovative approaches to manage multitenant platforms, underlying storage considerations, and database operations. They reveal the importance of Kubernetes features such as namespaces, storage orchestration, and policy enforcement in supporting a wide array of applications, including AI-intensive workloads. Through case studies and technical explanations, attendees learn about multitenancy architectures and strategies to leverage Kubernetes for data-intensive environments.

Interesting Product:

Virtual Clusters (vClusters) by Loft

VClusters provide a way to create lightweight virtual Kubernetes clusters within a single physical cluster. Each virtual cluster has its own API server, resource isolation, RBAC, and network policies. This allows for strong isolation between tenants.

Loft now offers vCluster pro that offers enterprise-grade features and support, including a management UI, SSO, audit logging, CoreDNS integration cross-cluster DNS, and running the vCluster control plane in its own dedicated Kubernetes cluster.

Observability and Performance Engineering

Instrumentation and performance optimization of applications are core aspects of this theme. Presenters discuss the use of flame graphs and profiling in OpenTelemetry to glean performance insights across diverse programming environments. Experiences in integrating profiling tools within the development ecosystem showcase how observability can lead to enhanced application tuning and optimization. The importance of understanding performance bottlenecks, across different stacks, and the shared learnings from these experiences underscore the ongoing pursuit of application efficiency and reliability in cloud native contexts.

Interesting Product

Grafana Cloud

Grafana Cloud and OpenTelemetry work together seamlessly to provide a complete observability solution.

Here’s how they integrate:

  1. Instrumentation: Use OpenTelemetry instrumentation libraries to generate metrics, logs, and traces from your applications and infrastructure.
  2. Data Collection: Use the OpenTelemetry Collector to receive, process, and export telemetry data to Grafana Cloud. Alternatively, you can use the Grafana Agent, which is a lightweight collector optimized for Grafana Cloud.
  3. Data Storage and Visualization: Grafana Cloud ingests and stores the telemetry data from OpenTelemetry, providing a unified platform for visualizing and analyzing metrics, logs, and traces using powerful Grafana dashboards.
  4. Application Observability: Grafana Cloud offers a dedicated “Application Observability” feature that provides pre-built dashboards and tools for monitoring applications instrumented with OpenTelemetry, following the Prometheus data model and semantic conventions.

By combining Grafana Cloud’s managed observability platform with OpenTelemetry’s vendor-neutral instrumentation and data collection capabilities, you can achieve end-to-end observability for your applications and infrastructure, without the overhead of managing and scaling the underlying components yourself.

Cloud Native Transformation in Telecommunications

As telecommunications companies contend with the burgeoning demands of 5G technology, adopting cloud native principles becomes increasingly critical. This theme covers the cloud native journey of telcos, emphasizing the collaborative ethos between service providers, vendors, and the broader cloud native community. Presentations and panels share experiences and challenges encountered along the path, offering lessons on digital transformation specific to telco applications and infrastructures. This sharing of insights across the ecosystem serves as a catalyst for innovation and adaptation within the dynamic telecom landscape.

Final Words

KubeCon Europe 2024 marked a significant milestone for platform engineering, showcasing it as the forefront of technological evolution within the cloud native ecosystem. Across 60 sessions, platform engineering was celebrated for its central role in enhancing the developer experience, with a focus on the implementation and optimization of IDPs. The event highlighted the intricate balance between development velocity and system reliability, underpinning the essence of platform engineering — creating toolchains and workflows that maximize developer productivity while ensuring the security and efficiency of data-driven applications.

The diversity of themes — from enhancing developer experience with IDPs, and managing data-centric workloads in Kubernetes, to leveraging AI for platform abstraction and improving cloud native networking with eBPF and Cilium — reflects the comprehensive approach needed to support the software development lifecycle. These discussions not only provided valuable insights into the current state of platform engineering but also pointed towards future trends, including the use of generative AI to ease developer burdens and the continuous evolution of platform abstractions to meet the needs of a growingly complex cloud native landscape.

As we reflect on the key themes of KubeCon Europe 2024, it’s clear that platform engineering is not just about tools and technologies; it’s about fostering a culture that values efficiency, security, and innovation. The event underscored the holistic view that platform engineering takes toward improving developer productivity and facilitating the creation of sophisticated, data-driven applications in a secure and efficient manner. With platform engineering at the helm, the future of cloud native development is poised for unprecedented growth and transformation, enabling organizations to harness the full potential of their technology investments. KubeCon Europe 2024 has indeed reaffirmed that in the ever-evolving landscape of cloud native technologies, platform engineering rules supreme.

Group Created with Sketch.
TNS owner Insight Partners is an investor in: PlanetScale, Octopus Deploy, Sysdig, Kubernetes.
TNS DAILY NEWSLETTER Receive a free roundup of the most recent TNS articles in your inbox each day.