Giant Conversations

Delve deep into IIoT as we discuss challenges, adoption, scaling, and cost benefits of IIoT Giant Conversations Episode #34 Topic: Manufacturing IIoT Hosted by: Tommy Hobin Special Guest: Manuel Gawert - Manufacturing IIoT - Why is IIoT Important? Introduction to Manuel Gawert: Technical PO for On-Premise Giant Swarm. Working with manufacturer's on Edge, IIoT, and On-Premise. Q: Why should a traditional manufacturer care about adopting IIoT now? Q: Can you give a real-world example /use case of how our IIoT platform helps improve factory operations? Manufacturing IIoT - Delving In Q: What's the "high-level" approach to scaling IIoT infrastructure across multiple facilities or geographic locations? Q: How do Kubernetes-native architecture benefit industrial customers compared to legacy systems? Q: How is IIoT important to digital transformation trends like AI, cloud adoption, or supply chain automation? Manufacturing IIoT at Giant Swarm Q: Do these platforms (our platform) help reduce costs? I'm sure they do, but how? Q: How do we handle interoperability with existing factory systems like SCADA, PLCs, and MES platforms? Q: Where do you see the future of manufacturing heading with platforms like ours? Manufacturing IIoT - From the Blog Blog Post: What works in Industrial IoT: lessons from real-world smart factories Blog Post: Unlocking smart factory success

Timo Derstappen, co-founder of Giant Swarm, shares his insights on how AI is revolutionizing the role of Platform Engineers. He discusses AI's dual nature, emphasizing the ‘garbage in, garbage out’ principle. Listeners will find intriguing points about the shift towards AI-driven product management and the integration of collaborative workflows in infrastructure development. Derstappen also addresses the challenges of maintaining human oversight amid rising automation, showcasing the importance of sustainable practices in the evolving tech landscape.

Dominik Schmidle, a Product Manager at Giant Swarm, dives into the world of cloud-native monitoring and observability. He explains why these concepts are essential for DevOps, shifting focus from traditional monitoring to a more holistic approach. The conversation touches on Grafana 12's exciting new features, including improved alerting and enhanced drill-down capabilities. Dominik also discusses the significance of OpenTelemetry in shaping observability tools and emphasizes the challenges of developing a comprehensive cloud-native observability platform.

In this episode, Oliver and Tommy discuss the Giant Swarm onsites, AI in the news and plug our NEXT event AI for Infrastructure Giant Conversations Episode #32 AI for Infrastructure Giant Swarm NEXT: Coffee edition with QAWare June 24th, at 9:30 am CEST Join Us: https://swarm.ly/3FnyeeL AI for Infrastructure: How Will AI Change the Life of a Platform Engineer? Episode Date: June 4, 2025 Topic: AI Event and Onsites + Swarmalicious News Hosted by: Oliver Thylmann Tommy Hobin This years Team Onsites Location: Mallorca - near Santanyi Sessions: Team Atlas and Cabbage: Release Requirements Session Release-as-a-Service Consistency across clusters(less exceptions) Testing as integral part of releases Be on a cluster fast(like weekly at least) Transparency for customers in a structured way Planeteers: (OLI FILLS IN) TeamUp: Improving feedback sessions Modular marketing services How we analyse conversion data Brainstorming using new methods: Rapid Ideation with AI Assistance Gifts and company SWAG (we are relaunching the Shop) Shield: Adopting Kubescape as our platform "core" (from Trivy) Kubescape would enable some features we don't currently have, but it is either missing or duplicating some we do have Example: Anomaly detection. Falco is also CNCF and has lots of users, so it's not clear whether we'd ditch it for Kubescape's agent. Are there things that even the expensive commercial tools don't do that we can do for customers? Examples: L1 security oncall - we take the pager for certain types of security issues and are the first entrypoint / triage for them Active remediation of findings - the scanner says "this is bad" and we make it good by actually working on customer applications Any fun stories? TeamUp got to run into 3 other teams while there - Teams Cabbage, Atlas, and Shield Jonas swam around 300 Meters to a deserted Island Jonas also snuck out at night with a night camara and came back with 2 sea cucumbers. He also mentioned that he scared some people We tested the well in the house - it works.. the water was probably not potable Next Onsite? Alicante - the whole company Swarmalicious AI News Mary Meeker's Bond AI Special Report: 340 slides into her new BOND “Trends – Artificial Intelligence” report. If you want to get the most out of it - ask ChatGPT to summarize it for you: Easy Takeaway - Adoption Facebook 4.5 years to reach 100M users ChatGPT 0.2 years Old AI News - Starting today [April 10th 2025], memory in ChatGPT can now reference all of your past chats to provide more personalized responses, drawing on your preferences and interests to make it even more helpful for writing, getting advice, learning, and beyond. Reference: Dress this dog in a pelican costume. Sign in the back written Half Bay Moon The previous memory feature—where the model would sometimes take notes on things I’d told it—still kept me in control. I could browse those notes at any time to see exactly what was being recorded, and delete the ones that weren’t helpful for my ongoing prompts. Yeti TikTok "powered" by AI: yetibubbavlog 21.2K Followers 175.4K Likes CloudFlare is trapping misbehaving bots in an AI Labyrinth. AI Labyrinth, a new mitigation approach that uses AI-generated content to slow down, confuse, and waste the resources of AI Crawlers and other bots that don’t respect “no crawl” directives. AI for Infrastructure AI for Infrastructure: How Will AI Change the Life of a Platform Engineer? AI-powered agents already making waves in the space — including tools like kagent, robusta.dev, and incident.io, which are helping teams detect issues faster, reduce toil, and boost operational confidence. Special Guests: Mario-Leander Reimer - Managing Director and CTO, QAware Timo Derstappen - CTO and Co-Founder, Giant Swarm Moderated by Tommy Hobin Next Weeks Podcast June 12, 2025 Dominik Schmidle will be talking about Monitoring and Observability.

We bring back Heather and Joe to talk (almost) all things London and KubeCon. Just as chaotic, fun, and educational as KubeCon itself. See you there! Giant Conversations Episode #31 Episode Date: March 28, 2025 Topic: KubeCon London Hosted by: Oliver Thylmann Tommy Hobin Special Guests: Heather Clancy Joe Salisbury ### See you at KubeCon - a marketer's take We'll be at KubeCon in London Drop by booth N450 Date: 1-4 April 2025 in London Giant Swarm Lineup - KubeCon: Marco Ebert: How To Gateway With Ingress - 140 Days InGate Giant Swarm Lineup - KubeCon (Platform Engineering Day): Dominik Kress: Platform-as-a-Product: First Insights From CNCF Platform Working Group's Ongoing Research Giant Swarm Lineup at Rejekts: Marcus Noble: Pod Deep Dive: Everything You Didn't Know You Needed to Know Joe Salisbury: The Cluster API Migration Retrospective: Live migrating hundreds of clusters to Cluster API Anyone on this Pocast a bit nervous? ### Stuff to Know Download the CityMapper app to get your way around London's Tube System The station near the venue is Custom House (there exit ExCel), available from Heathrow Terminals by the Elizabeth line What parties are you attending? ### Fun and a little nerdy bit of things to do in London Only a 5 minute walk from Holborn Tube Station: The Novelty Automation Museum - very reminiscent of our invention hero Simone Giertz Oi, check this out >> https://www.timhunkin.com/a260_travel-by-tube.htm Have y'all thought about going to an arcade? Do y'all say y'all? I don't care, here's a list of arcades. Horizon 22! Natural History Museum (South kensington in general - Science Museum, V&A, kensington creperie) spitalfields market (more food) sky garden (alternative to horizon22 hms belfast (if you're into navy stuff) london transport museum (if you're into trains lol) tooting market / broadway market / the wheatshef shout out to the ends Hayward Gallery Food Guide The Dishoom paradox (Kricket) Blacklock (lol at Joe) Gaucho Borough Market Bermondsey Beer Mile 40 Maltby Street. ### Talks at Rejekts/KubeCon that peak our interest (KC) Marco Ebert: How To Gateway With Ingress - 140 Days InGate (R) Marcus Noble: Pod Deep Dive: Everything You Didn't Know You Needed to Know (R) Joe Salisbury: The Cluster API Migration Retrospective: Live migrating hundreds of clusters to Cluster API (KC) Dominik Kress: Platform-as-a-Product: First Insights From CNCF Platform Working Group's Ongoing Research (R) James Laverack, Luigi Zhou: High-Scale Networking for ML Workloads With Cilium (R) Frederic Branczyk, Jimmy Zelinskie: From Milliseconds to Microseconds: Pushing Kubernetes Workloads to the Limit (KC) Joe's Wildcard: Dennis Marttinen, Security and Cloud Computing (SECCLO) Master Student, Aalto University: Thousands of Virtual Kubelets: 1-to-1 Mapping a Supercomputer To Kubernetes With Supernetes

It's not a cut together sho of memories. Tech news keeps coming in and we're here to decipher it for you... still. Giant Conversations Episode #30 Episode Date: March 12 2025 Hosted by: Oliver Thylmann Tommy Hobin Contributors: Puja Abbassi Martin Proffitt Marcus Noble Marian Steinbach ### Swarmalicious News We'll be at KubeCon in London Drop by booth N450 Date: 1-4 April 2025 in London Giant Swarm Lineup - KubeCon: Marco Ebert: How To Gateway With Ingress - 140 Days InGate Giant Swarm Lineup at Rejekts: Marcus Noble: Pod Deep Dive: Everything You Didn't Know You Needed to Know Joe Salisbury: The Cluster API Migration Retrospective: Live migrating hundreds of clusters to Cluster API Oliver's Take: You don't often go to conferences. I've heard there are three kinds of people who go: Heavy learners who can't miss a session, take notes, and bring back boatloads of informatio Those who just kind of hang around and network a bit Those who party till the daytime and go to the conference slightly drunk …. there's proibably a fourth. The Evolution of IT Operations and Opsgenie: Opsgenie end of support – effective April 5th, 2027: On this date, access to Opsgenie will be shut off, and the product will no longer accessible. Any unmigrated data in Opsgenie will also be deleted at this point. This a reasonable amount of time to do something about it - right? Undocumented commands found in Bluetooth chip used by a billion devices Researchers have identified undocumented vendor-specific commands in the ESP32 microcontroller, a widely used chip enabling Wi-Fi and Bluetooth connectivity in over a billion IoT devices as of 2023. The security issues include: Spoof Trusted Devices: Impersonate legitimate devices to gain unauthorized access. Access Data Illegally: Retrieve sensitive information without permission. Network Pivoting: Move laterally within a network to compromise additional devices. Establish Persistence: Maintain long-term unauthorized access to devices. On May 5, Microsoft’s Skype will shut down for good: Moving to Teams… Why is Microsoft doing this? Integration Challenges: After acquiring Skype in 2011, Microsoft faced difficulties integrating it into its ecosystem, leading to technical issues and user dissatisfaction. Rise of Competitors: The emergence of platforms like Zoom and WhatsApp, offering user-friendly interfaces and reliable services, diminished Skype's market share. Strategic Shift to Teams: Microsoft's focus shifted to Teams, a platform that seamlessly integrates with its suite of productivity tools, aligning better with modern communication needs. Cast AI 2025 Kubernetes Cost Benchmark Report Key findings: Only 10% of provisioned CPUs are utilized Memory utilization averages just 23% The gap between provisioned and requested resources is massive— 40% for CPU, 57% for memory The data for our 106 clusters shows CPU Utalization Mean between 9.6% and 13.7%, Max 19.6% Memory Utalization Mean between 20.3% and 40.9%, Max 46.7% How data centres can chase renewable energy across Europe In Ireland, data centers' share of electricity consumption was 21% in 2023 and is projected to reach 32% by 2026. This article covers Geographical Flexibility: Locating data centers in regions abundant in renewable resources—such as wind-rich northern Europe or sun-rich southern Europe—allows for optimal utilization of local renewable energy capacities. Operational Flexibility: Scheduling non-time-sensitive computational tasks during periods of high renewable energy availability can reduce reliance on non-renewable power sources. Dynamic Load Shifting: By distributing computing tasks across various locations based on real-time renewable energy availability, data centers can minimize energy costs and emissions. For instance, when wind generation is low in one area, tasks can be shifted to another region with higher wind output. The European Centre for Medium-Range Weather Forecasts (ECMWF) has introduced the Artificial Intelligence Forecasting System (AIFS), an AI-based weather model that significantly enhances forecasting accuracy while reducing energy consumption by approximately 1,000 times compared to traditional methods. Key Features of AIFS: Enhanced Accuracy: AIFS leverages machine learning techniques to provide more precise weather predictions, improving the reliability of forecasts. Energy Efficiency: The system's AI-driven approach drastically reduces the computational resources required, leading to substantial energy savings. Operational Integration: As the first operational forecasting model extensively utilizing machine learning across various parameters, AIFS represents a significant advancement in meteorological practices.

Confidential computing, Flux 2.5 GA, agent mode for GitHub Copilot, and much more. Giant Conversations Episode #29 Episode Date: March 5 2025 Topic: Puja Two Y'all! Hosted by: Oliver Thylmann Tommy Hobin Special Guest: Puja Abbassi Contributors: Puja Abbassi Timo Derstappen Simon Weald Zach Stone Martin Proffitt Marcus Noble Subscribe to Marcus' Kubernetes Newsletter https://cloudnative.now/ ### Swarmalicious News Marcus Noble: Tekton Incubation Application - This is from May 1, 2024. Is it new? Martin Proffitt: Introducing Flux 2.5 GA. General Availability of the Bucket API: The Bucket API has reached General Availability (GA), providing a stable interface for managing bucket resources. CLI Enhancements: The flux command-line interface now includes a create secret proxy command, facilitating the creation of proxy secrets. Additionally, the create source commands have been updated with a --proxy-secret-ref flag to support proxy configurations. Dependency Updates: Flux components have been updated to align with Kubernetes v1.31.1, ensuring compatibility and leveraging the latest Kubernetes features. Testing Improvements: Conformance tests now cover Kubernetes versions 1.29 through 1.31, ensuring robust support across these versions. Puja Abbassi: Six Sins of Platform Teams where Sergey Tselovalnikov discusses common pitfalls that platform teams encounter, such as: Structuring teams around specific solutions Losing empathy towards product engineers Overengineering solutions Neglecting user feedback Imposing mandatory usage Lack of clear success metrics Marcus Noble: CVE-2025-0426: Node Denial of Service via kubelet Checkpoint API. A security issue was discovered in Kubernetes where a large number of container checkpoint requests made to the unauthenticated kubelet read-only HTTP endpoint may cause a Node Denial of Service by filling the Node's disk. What is this process? Simon Weald: Tolerating full cloud outages with Monzo Stand-in Monzo introduces "Monzo Stand-in," a backup banking infrastructure hosted on Google Cloud Platform (GCP) designed to maintain essential banking services during outages of their primary platform on Amazon Web Services (AWS). Notes: Independent Operation: Monzo Stand-in runs on GCP, while the primary platform operates on AWS, ensuring that a failure in one doesn't affect the other. Essential Services Coverage: During primary platform outages, Stand-in supports vital functions such as card spending, cash withdrawals, bank transfers, balance checks, transaction history, and card freezing/unfreezing. Simplified App Interface: When Stand-in is active, the Monzo app automatically transitions to a streamlined user interface, focusing on essential banking features to maintain user accessibility. Risk Mitigation Strategy: By developing a separate set of services for Stand-in, Monzo avoids complexities and potential issues related to real-time data replication and consistency between platforms. Puja Abbassi: Confidential computing is a new technology that enables 1Password to bring its end-to-end encryption model into the cloud. Confidential AI with: https://www.edgeless.systems/ ### AI News Puja / Timo: Introducing agent mode for GitHub Copilot in VS Code, announcing the general availability of Copilot Edits, and providing a first look at our SWE agent. Puja: ChatGPT for DevOps: otto8.ai. Easily integrate CLIs, Docker Images, JavaScript, Python, and Shell scripts with AI.

Our colleague Puja Abbassi joins us and we discuss some of our BFFs content. We're talk about you Frederic, Viktor, Tim, and Millennium BCP. Giant Conversations Episode #28 Episode Date: February 12 2025 Topic: Hosted by: Oliver Thylmann Tommy Hobin Special Guest: Puja Abbassi Contributors: Puja Abbassi Timo Derstappen Simon Weald Zach Stone Martin Proffitt Marcus Noble Xavier Avrillier ### Swarmalicious News Zach Stone - TIL about this k8s native global load balancer project, currently CNCF sandbox: A cloud native Kubernetes Global Balancer Advantages: Kubernetes Native Integration High Availability and Automatic Failover Flexible Load Balancing Strategies New Applications: Multi-Cluster Deployments Integration with Liqo Enhanced Observability Puja Abbassi - Our friends from Polar Signals built a nice tool to detect cross-zone traffic: eBPF Tool Identifies Cross-Zone Kubernetes Network Traffic From The New Stack: Polar Signals couldn't figure out why it had so much expensive cross-zone traffic. So it turned to eBPF to deliver the answers, and cut its cloud bill in half. Martin Proffitt - Why I Changed My Mind About Backstage? A Brutally Honest Opinion What is his brutally honest opinion? It IS difficult to work with. But so was Kubernetes in the past. Better options back then would have been Docker, Mesosphere, etc. But NOW K8s is the defacto standard. Port for example may be easier now, but he believes Backstage will become the main choice in the future. Predominantly because of CNCF involvement. Marcus Noble - Kubernetes History Inspector This tool parses Kubernetes and Google Kubernetes Engine (GKE) logs, generating a comprehensive timeline of all events within a cluster. Key Features: Comprehensive Event Timeline Information-Dense Interface Xavier Avrillier - Cilium Releases v1.17.0: https://github.com/cilium/cilium/releases/tag/v1.17.0 The new Cilium version includes some pretty cool new features including: Quality of Service: Annotate your Pods for Guaranteed, Burstable or BestEffort egress network traffic priority FQDN Filtering for hostNetwork: Use CiliumClusterwideNetworkPolicy to configure Layer 7 filtering for DNS requests on nodes in the cluster AI News Timo Derstappen: - Let's different llms talk to each other to find consensus. llm-consortium Key Features: Multi-Model Orchestration: The system coordinates responses from multiple LLMs simultaneously, allowing each model to contribute its unique strengths to the problem-solving process. Iterative Refinement: Through multiple rounds of evaluation, the consortium refines responses to improve accuracy and coherence, ensuring that the final output is well-vetted. Advanced Arbitration: A designated arbiter model synthesizes responses from various LLMs, evaluating and combining them to produce optimal results. Blockchain Verification: Decisions and consensus are recorded on the Solana blockchain, providing transparency and verifiability of AI-generated outputs. Puja Abbassi - DeepSeek claims its ‘reasoning’ model beats OpenAI’s o1 on certain benchmarks Chinese AI lab DeepSeek has released an open version of DeepSeek-R1, its so-called reasoning model, that it claims performs as well as OpenAI’s o1 on certain AI benchmarks. Tim from Netlight mentioned is running most of his LLM workloads on K8s through: https://github.com/vllm-project/vllm It's a bit complex, but it supposedly what OpenAI also uses under the hood and very very powerful: DOCS Timo Derstappen - Openai launched their first agent: https://www.youtube.com/live/CSE77wAdDLg Simon Weald - Trolling SMS spammers with Ollama: https://evan.widloski.com/software/sms_llm/

Oliver and Tommy delve into his Cloud Native predictions for 20 25 and recap his predictions from 2024 And looking into the future from the Team at Giant Swarm Giant Conversations Episode #27 Topic Cloud Native Predictions for 2025 from Giant Swarm Hosted by: Tommy Hobin Oliver Thylmann Last year's predictions: Was Oli right? Decision making will be a people topic CNCF will gain in strength and add key projects CNCF Project List and CNCF Project Count More companies or projects will fold or merge The IDP, Internal Developer Platform trend will go full force GitOps becomes the standard Edge clusters gain traction AI Ops is a thing - Community forks for Open Source projects on the horizon Main topic: Cloud native predictions for 2025 Looking Ahead: 2025's Unfolding Story Predictions gathered from Swarmalicious: 1. Gateway API: The release of Gateway API v1.2.0 in November 2024 introduced features such as WebSocket support, timeouts, and retries The upcoming Gateway API v1.3 is expected to be released around the end of January 2025. It will possibly include: Expanded Protocol Support Enhanced Configuration Options Improved Observability Authentication and Security Enhancements Better Cross-Cluster and Multi-Tenancy Support User experience and usability (UI) 2. Stronger focus on UI: Flux, Headlamp, Backstage etc.: Why Together They Matter and complement each other: Flux ensures operational consistency through GitOps. Headlamp simplifies operational management for Kubernetes clusters. Backstage abstracts Kubernetes details to enhance developer productivity. This will help solve: Developer and Operator Alignment: Together, they bridge the gap between infrastructure operators and application developers. Improved Kubernetes Adoption: By addressing different pain points, these tools lower the barrier for Kubernetes adoption across organizations. 3. More indie community activity. KCDs and more "grassroots" style events and communities (the CNCF is still very active in these): DevOpsDays KCDs CloudNativeDay events Industry-Specific Kubernetes Communities Financial Services: FinOps Foundation, Cloud Native Financial Services Day (CNCF) Healthcare and Life Sciences: HealthTech User Groups (Focus on HIPPA Compliance and Security) Telecommunications: Open Networking Foundation, Kubernetes in Telco (CNCF SIG) Gaming: Game Developers Kubernetes Group, Open GameStack Initiative Manufacturing: CNCF IoT Edge Working Group, Industry 4.0 Kubernetes Group Retail and E-Commerce: Retail Kubernetes Users Group, CNCF App Delivery SIG Automotive: Autoware Foundation, CNCF Edge Computing SIG 4. Formalisation in the platform space around capabilities and maturity: This formalization is driven by the need to manage complexity, foster innovation, and ensure scalable, secure, and reliable platform ecosystems. Platform Capability Frameworks Maturity Models Governance and Compliance Metrics for Measuring Maturity Collaboration Between Teams Adoption of Standards Emphasis on Innovation and Adaptability

Or Minecraft > eBPF Giant Conversations Episode #26 Topic Minecraft / Kubernetes / eBPF Hosted by: Tommy Hobin Joseph Salisbury Starring: James Laverack Joseph Salisbury Main topic: Interviewing James Laverack Q&A with James Laverack and Joe Salisbury - more Minecraft than anything else References to Minecraft Mods: https://modrinth.com/ mrpack container: https://github.com/JamesLaverack/mrpack-container Theme: Talks and Topics “Building an Application-Specific Container Image Builder” (Minecraft) Theme: Software and Passion Software you love and loath (may be the same) James on the Mic: Building an Application-Specific Container Image Builder | James Laverack - Rejekts NA 2024 How eBPF Actually Works: Cloud Native Rejekts) - Rejekts EU 2024 Talking to Kubernetes with Rust - James Laverack - KubeCon EU 2023 In Clouds We Trust | James Laverack & Josh Van Leeuween) - KCDUK 2022 Kubernetes 1.24, with James Laverack - Podcast 2022 I like this guy: Meet James