Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Serverless Computing in Azure Kubernetes Service mit KEDA

6848c06ef647ab606c668cc5264c0fc9?s=47 Thorsten Hans
November 04, 2020

Serverless Computing in Azure Kubernetes Service mit KEDA

Slides from my Thinktecture Webinar on serverless computing in Azure Kubernetes Service (AKS) with KEDA.

6848c06ef647ab606c668cc5264c0fc9?s=128

Thorsten Hans

November 04, 2020
Tweet

Transcript

  1. Serverless Computing in Azure Kubernetes Service mit KEDA Thorsten Hans

    @ThorstenHans Consultant
  2. Focus: Cloud Native & Infrastructure as Code Consultant @ Thinktecture

    AG thorsten.hans@thinktecture.com thorsten.hans@gmail.com thinktecture.com thorsten-hans.com @ThorstenHans Thorsten Hans
  3. Talking Points Scaling in Kubernetes Scaling with KEDA Hands-On

  4. Kubernetes scaling capabilities § Manual scaling § Load based scaling

    § Horizontal Pod Autoscaler § AKS Cluster Autoscaling Scaling in Kubernetes
  5. Manual scaling Scaling in Kubernetes

  6. Horizontal Pod Autoscaler (HPA) Scaling in Kubernetes

  7. Cluster Auto-Scaler in Azure Kubernetes Service (AKS) Scaling in Kubernetes

  8. Talking Points Scaling in Kubernetes Scaling with KEDA Hands-On

  9. Kubernetes Event-driven Autoscaling § https://keda.sh § Cloud Native Computing Foundation

    sandbox project § Vendor Agnostic § Contextual Autoscaling § Ships with 25 built-in scalers § Built-in Azure Functions Support What is KEDA
  10. Architecture KEDA https://keda.sh/docs/2.0/concepts/

  11. Messaging System / Database / Cache / … Architecture Simplified

    KEDA Kubernetes Cluster KEDA Operator Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod Kubernetes Pod
  12. Flexbile scaling § Scale based on actual load § Built-in

    scalers for § Databases § Messaging engines § Telemetry systems § Cache engines § Unified configuration approach § ScaledObject specification KEDA Scalers
  13. Talking Points Scaling in Kubernetes Scaling with KEDA Hands-On

  14. Async Processing with Azure Functions and KEDA Demo Architecture Local

    Machine Message Producer Azure Public Cloud Azure Storage Account Queue (aaa-tasks) Azure Storage Account Queue (aaa-transformed) Azure Storage Account Blobs (processed) Azure Kubernetes Service KEDA Operator Message Transformer (Azure Functions Container) Message Dispatcher (Azure Functions Container) Horizontal Pod Autoscaler
  15. Kubernetes Event-driven Autoscaling - Install KEDA - Deploy Azure Functions

    - Configure Connections to Azure Storage - Deploy ScaledObjects - Produce Load to verify scaling Demo
  16. KEDA outperforms traditional resource based scaling in Kubernetes where fine-granular

    components have to scale contextually Takeaway
  17. DEMO Source Code is on GitHub https://github.com/thinktecture/keda-webinar-november-2020