Lightnews — Scholar-powered news

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8h

Enter llm-d new open source tool/approach treat the LLM and its runtime as disaggregated, first-class components inside K8s. Break the container into parts (cache, prefill/decode, GPU-bound work, CPU-bound work) and let the platform place and scale each piece independently.
dev.to/saimsafdar/i...

Introduction to llm-d Open-source Kubernetes-native Framework for Distributed LLM Inference | Ep 140 #cloudnativefm

I recently had a great conversation with the Red Hat team about llm_d, a new open-source effort...

dev.to

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8h

How would you explain this to someone who’s just starting out as a Platform Engineer?

2

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 10h

LLMs are monoliths, which can be a major cause for your CPU/GPU compute bills 📈. What if we can build a K8s-native distributed inference stack that brings cache-aware routing and disaggregated serving to LLMs?

Weclome LLM-D which does that, make ur compute bills 📉.
www.youtube.com/shorts/rI8zF...

Ep 140 Shorts: Introduction to llm-d Open-source K8s-native Framework for Distributed LLM Inference

YouTube video by Cloud Native Podcast

www.youtube.com

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 1d

Platform Engineering initiatives are shifting from theory to practice, but what actually works? Luckily, now we have the data to find patterns. Thanks to @OctopusDeploy platform Engineering Pulse Report, luckily enough to be a contributor alongside other industry experts.
youtube.com/shorts/5NbSn...

#cloudnativewisdom010 Reviewing: Octopus Deploy Platform Engineering Pulse Report 2025

YouTube video by Cloud Native Podcast

youtube.com

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 2d

Happy birthday, my best friend @virtualized6ix.wtf, hope you've great day, I'm hoping for a day I can wish you inperson, sending alot of 💙 + 🤗

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 3d

llm-d is a new opensource tool and approach designed to make serving generative models on K8s efficient, scalable, and cost-effective by introducing cache-aware routing, disaggregated serving (pre-fill/decode), and K8s-native scheduling & gateways.

🎧 to #CloudNativeFM 👇 youtu.be/2Wtug1kTwUk

Introduction to llm-d Open-source Kubernetes-native Framework for Distributed LLM Inference | Ep 140

YouTube video by Cloud Native Podcast

youtu.be

1

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 4d

With current AI agentic hype, we are already in this phase, just won't realize it.

1 3

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 5d

This is your reminder that we are living in a world where Windows runs Linux, that's what we want.

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 5d

Thanks to @cedricclyburn.com & Christopher Nuland, I'm pretty much impressed with llm-d.ai: a Kubernetes-native distributed inference stack that brings cache-aware routing and disaggregated serving to LLMs.

Watch the full episode to know more: youtu.be/2Wtug1kTwUk

1 1 1

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 6d

Episode 3: Golden Paths vs Developer Freedom just dropped

Golden paths must be contextual (co-design w/devs, prioritize feedback loops), stricter when regulation or scale demand it, lighter where speed and experimentation are business priorities.

Watch the full episode: youtu.be/_FPdHexAyfk

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 7d

I remember a time when my dev environment for day in and day out was just React, since I've moved to DevOps, I haven't seen this space closely.

React & React Native are now transitioning to the React Foundation under the Linux Foundation

What are your thoughts?
engineering.fb.com/2025/10/07/o...

Introducing the React Foundation: The New Home for React & React Native

Meta open-sourced React over a decade ago to help developers build better user experiences. Since then, React has grown into one of the world’s most popular open source projects, powering over 50 m…

engineering.fb.com

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 7d

Congrats to all the maintainers/contributors #knative project, Welcome to the graduation club.

I've high hopes that recognition reflects a mature ecosystem of primitives, an even more compelling option for platform teams building serverless & event-driven platforms.
www.cncf.io/announcement...

Cloud Native Computing Foundation Announces Knative’s Graduation

Graduation marks Knative’s readiness for widespread production use, with upcoming features aimed at bridging legacy systems and expanding AI and cloud native integrations Key Highlights: SAN FRANCISCO...

www.cncf.io

3

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 7d

The 𝘾𝙤𝙙𝙚 𝙏𝙤 𝘾𝙪𝙡𝙩𝙪𝙧𝙚: 𝙋𝙡𝙖𝙩𝙛𝙤𝙧𝙢 𝙀𝙣𝙜𝙞𝙣𝙚𝙚𝙧𝙞𝙣𝙜 𝙐𝙣𝙥𝙖𝙘𝙠𝙚𝙙 𝙁𝙤𝙧 𝙀𝙣𝙩𝙚𝙧𝙥𝙧𝙞𝙨𝙚𝙨 series is well underway on both the #CloudTherapist and #cloudnativefm YouTube channels.

Ep 1 – www.youtube.com/watch?v=Tgh9...

Ep 2 – www.youtube.com/watch?v=QxUM...

EP 3 – youtu.be/_FPdHexAyfk

1

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8d

#nobelprize2025 goes where it "belongs", All of this hard work will eventually give us a "quantum computer" way earlier than everyone can predict. In 2026, I'll be looking up this space more closely because we can solve problems; previously, we were not sure how, but now we can reimagine.

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8d

Code to Culture panel series

Episode 3: Golden Paths vs Developer Freedom #cloudnativefm

Golden paths must be contextual (co-design w/devs, prioritize feedback loops) stricter when regulation or scale demand it, lighter where speed & experimentation are business priorities

-> youtu.be/_FPdHexAyfk

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8d

Code to Culture: Platform Engineering panel series

Episode 3: Golden Paths vs Developer Freedom just dropped.

Watch it here: youtu.be/_FPdHexAyfk

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 8d

Happy Tuesday!!!

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 9d

Code to Culture: Platform Engineering panel series

What if Devs don't want to do PE? What if they prefer their own freedom? What if some of the tools we choose in our "golden path" are NOT what Devs like??

Ep3: Dev Freedom vs Golden Paths dropping soon: youtube.com/@cloudnativefm

1

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 9d

Why, this is so true!!!

2

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 9d

Monday vibes!!! Hope you've a great day.

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 11d

In this episode, we speak with Qasim Sarfraz (Inspektor-Gadget maintainer) to answer a common question: If I already have Prometheus, Grafana, and OpenTelemetry, do I still need Inspektor-Gadget?

Watch #CloudNativeWisdom: youtu.be/9ZZcXeUt4A8

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 12d

Can Inspektor-Gadget handle K8s autoscaling needs alone?

Short answer: Not a standalone autoscaler, but a powerful data collection for observability (Prometheus/OpenTelemetry), combining those signals with HPA/VPA, KEDA, to achieve production-grade autoscaling behavior.

-> youtu.be/EsmYRDPTXRc

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 13d

Reminder: True success comes with "What People Don't SEE".

1

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 13d

Perforator connects to real profiling data from Grafana Pyroscope (CPU & memory usage), That means you can immediately see “hot spots” – the parts of your code that are using the most resources – without switching to external dashboards.

A deep dive into Perforator -> youtu.be/gJSEmjDrD4E

cloudnativeboy.bsky.social @cloudnativeboy.bsky.social · 13d

That time of the year when you realize "Me and My 2025", is it just me or anyone else feeling the same.

2