Mastering Kubernetes Resiliency and Costs with Dynatrace

#Kubernetes has become the go-to platform for deploying applications in recent years. And while it offers a multitude of impressive features, it also comes with a set of responsibilities and potential challenges – specifically around running scalable and resilient workloads without losing track of costs.

In this Observability Clinic we have Henrik Rexed, Cloud Native Advocate at #Dynatrace and host of #IsItObservable, giving us insights with live demos on:

Measuring the Cloud Costs of your K8s Clusters
The Impact of non-optimized K8s resource allocations
The latest K8s features in Dynatrace
How to leverage HPA (Horizontal Pod Autoscaler) with any metric from Dynatrace

Henrik also educates on how to best observe and monitor your K8s workloads, covers essential metric best practices, and how to identify side effects of improper auto-scaling.

Here are the links as discussed in this clinic:

Chapter List:
00:00 - Introduction
00:50 - What you're going to learn today
03:00 - Once Upon a Time (eCommerce Story)
04:10 - The cost of a cluster
04:30 - Everything has a price
05:53 - Optimizing the Cost of a Cluster
06:42 - Resize your Workload Size
08:38 - Overview Demo #1
10:17 - Live Demo #1
14:38 - Optimization Results
18:31 - Request Limits Explained
22:10 - Dynatrace Out-of-the-Box K8s Alerting
25:52 - K8s Observability with Dynatrace Explained
28:21 - Live Demo #2 - Dynatrace K8s Settings
33:08 - Auto-Scaling your K8s Cluster
34:22 - HPA (Horizontal Pod Autoscaler)
35:40 - Overview Demo #2 - HPA
36:25 - Live Demo #2 - HPA
38:58 - HPA Scaling Results Explained
40:00 - Better Scaling with Keptn Metrics Server
42:50 - Live Demo #3 - Scaling based on Throttling
44:40 - HPA Scaling Results based on Throttling
45:59 - Take Aways
48:47 - Open Q&A

Stay up-to-date with Dynatrace! Follow us on:

[roberto_camp]

Really nice demo. I notice the github repo is GKE-oriented. Would you happen to have an EKS version of the code? Or would all the code basically work the same way on an EKS cluster?