From 97735bafa8b3f6aee9fa205b95e126b39242a434 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=B4me=20Petazzoni?= Date: Fri, 5 Nov 2021 14:19:28 +0100 Subject: [PATCH] =?UTF-8?q?=E2=9E=95=20Mention=20KEDA?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- slides/k8s/hpa-v2.md | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/slides/k8s/hpa-v2.md b/slides/k8s/hpa-v2.md index 588ce936..e6c06815 100644 --- a/slides/k8s/hpa-v2.md +++ b/slides/k8s/hpa-v2.md @@ -652,6 +652,22 @@ Also check the logs of the `prometheus-adapter` and the `kube-controller-manager - [Prometheus adapter configuration walkthrough](https://github.com/DirectXMan12/k8s-prometheus-adapter/blob/master/docs/config-walkthrough.md) +--- + +## Discussion + +- This system works great if we have a single, centralized metrics system + + (and the corresponding "adapter" to expose these metrics through the Kubernetes API) + +- If we have metrics in multiple places, we must aggregate them + + (good news: Prometheus has exporters for almost everything!) + +- It is complex and has a steep learning curve + +- Another approach is [KEDA](https://keda.sh/) + ??? :EN:- Autoscaling with custom metrics