From 84abb7e1b9f7bfd7fca61084cbfaaa137d9be640 Mon Sep 17 00:00:00 2001 From: Nicolai Ort Date: Tue, 22 Jul 2025 11:19:22 +0200 Subject: [PATCH] docs(day2): Added edge cpu talk notes --- content/_index.md | 9 ++++-- content/day2/03_k3s-gpu.md | 56 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 63 insertions(+), 2 deletions(-) create mode 100644 content/day2/03_k3s-gpu.md diff --git a/content/_index.md b/content/_index.md index 9504529..3516c55 100644 --- a/content/_index.md +++ b/content/_index.md @@ -5,12 +5,17 @@ title: Cloud Native Summit Munich 2025 All about the things I did and sessions I attended at Cloud Native Summit 2025 in Munich. -This current version is probably full of typos - will fix later. This is what typing the notes blindly in real time get's you. +This current version is probably full of typos - might fix later (prbly won't tbh). This is what typing the notes blindly in real time get's you. ## How did I get there? I attended Cloud Native Rejekts and KubeCon + CloudNativeCon Europe 2025 in London and some of the atendees reccomended checking out CNS Munich for another event in the same spirit as Cloud Native Rejekts. -After a short talk with my boss, I there by my employer [DATEV eG](https://datev.de) alongside two of my coworkers. +After a short talk with my boss, I got sent there by my employer [DATEV eG](https://datev.de) alongside two of my coworkers. + +## And how was it. + +I'd say that attending CNS Munich 2025 was worth it. The event is pretty close to my place of employment (2hrs by car or train) and relatively small in size (400 attendees). The talks varied a bit - the first day had a bunch of interesting talks but the second day indulged in ai-related talks (and they were not quite my cup of tea). This might me fine for others but I've heard enogh about ai use cases for the coming years at the last events i attended (or just reddit). +Maybe disributing the ai-talks over the two days - while always providing an interesting alternative - might be the right move for next time. ## And how does this website get it's content diff --git a/content/day2/03_k3s-gpu.md b/content/day2/03_k3s-gpu.md new file mode 100644 index 0000000..3c06622 --- /dev/null +++ b/content/day2/03_k3s-gpu.md @@ -0,0 +1,56 @@ +--- +title: "Brains on the edge - running ai workloads with k3s and gpu nodes" +weight: 3 +tags: + - ai + - gpu +--- + + + + +I decided not to note down the usual "typical challenges on the edge" slides (about 10 mins of the talk) + +## Baseline + +- Edge can be split up: Near Edge, Far Edge, Device Edge +- They use k3s for all edge clusters + +## Prerequisites + +- Software: GPU Driver, Container Toolkit, Device Plugin +- Hardware: NVIDIA GPU with a supported distro +- Runtime: Not all runtimes support GPUs (containerd and CRI-O do) + +## Architecture + +```mermaid +graph LR +subgraph Edge + MQTT + Kafka + Analytics + + MQTT-->|Publish collected sensor data|Kafka + Kafka-->|Provide data to run|Analytics +end +subgraph Azure + Storage + Monitoring + MLFlow + + Storage-->|Provide long term analytics|MLFlow +end + +Analytics<-->|Sync models|MLFlow +Kafka-->|Save to long term|Storage +Monitoring-.->|Observe|Storage +Monitoring-.->|Observe|MLFlow +``` + +## Q&A + +- Did you use the nvidia gpu operator: Yes +- Which runtime did you use: ContainerD via K3S +- Why k3s over k0s: Because we used it +- Were you power limited: Nope, the edge was on a large ship