BACKLOG.MD

Backlog

Edge scope

Light Event Bus

• Replace Mosquito by Microsoft E4K as the light MQTT message broker: E4K will provide much better integration with Azure management, security, etc.

Security

• Authentication/Authorization: Authentication/Authorization based on an Identity Server such as Keycloak or Duende (aka.IdentityServer4)

Automated provisioning of cameras/devices

• Automated provisioning of cameras/devices with AKRI
  • https://github.com/project-akri/akri

Custom code and Kubernetes

• Advanced Versioning of Microservices (semantic): Implement a more advanced versioning system based on aspnet-api-versioning or comparable system: https://github.com/Microsoft/aspnet-api-versioning
• Migrate K8s deployment YAML to HELM charts: Related to Open Application Model OAM which uses HELM.
• Refactor to comply with "Project Radius" and/or Open Application Model OAM to have the app definition decoupled from app deployment on multiple supported environments.
  • OAM:
    • https://oam.dev/
    • https://github.com/oam-dev
    • https://cloudblogs.microsoft.com/opensource/2019/10/16/announcing-open-application-model/
    • https://learn.microsoft.com/en-us/shows/azure-friday/intro-to-rudr-a-kubernetes-implementation-of-the-open-application-model
    • https://github.com/kubevela/kubevela
• Implement a "Zipkin enable/diable" mechanism: Need a simple flag to enable/disable Zipkin, since in "production" it can impact perf.
• High coverage of Unit Tests and Functional Tests
• Local MEC error loging
• SPA web app with related API Gateway: Such as a .NET BLAZOR WebAssembly SPA and Envoy or OSM as API Gateway.
• Shared folder in Kubernetes for images/video files: Instead of embeded/serialized images in JSON and MongoDB database.
• NGINX ingress in Kubernetes (if needed)
• Support of shared GPU across multiple pods in Kubernetes: When supported by K8s, needed changes in implementation for inferencing the AI models with GPU.
• ACR & MAR vs. Docker Hub: Evaluate if we'll use ACR or MAR as Docker Registry instead of Docker Hub. TBD.

Deployment

• Flux and GitOps: Evaluate Flux and GitOps for deployment of multiple sites and multiple Kubenetes config in sync. Flux is a tool for keeping Kubernetes clusters in sync with sources of configuration (like Git repositories), and automating updates to configuration when there is new code to deploy.
  • https://fluxcd.io/flux/
  • https://learn.microsoft.com/en-us/azure/azure-arc/kubernetes/tutorial-use-gitops-flux2?tabs=azure-cli
  • https://learn.microsoft.com/en-us/azure/azure-arc/kubernetes/conceptual-gitops-flux2

Alerts Rule Engine

• Comprehensive Alerts Rule Engine based on industry/OSS solutions such as Kuiper, or Neuron

AI / ML / DeepLearning

• Cascade Models support
• Shared GPU across K8s Pods: Use NVIDIA DeepStream to potentially share GPUs for multiple Kuberentes Pods. References:
  • NVIDIA GPU sharing in Kubernetes: https://developer.nvidia.com/blog/improving-gpu-utilization-in-kubernetes/
  • GPU time slicing in K8s may help to have pods share GPUs: https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/gpu-sharing.html
  • Azure DeepStream architecture, to compare code using NVIDIA DeepStream from Azure IoT Edge: https://github.com/microsoft/AzureDeepStreamAccelerator#azure-deepstream-accelerator-architecture
  • https://blog.ml6.eu/a-guide-to-gpu-sharing-on-top-of-kubernetes-6097935ababf
  • Schedule GPUs in Kubernetes: https://kubernetes.io/docs/tasks/manage-gpus/scheduling-gpus/

Cloud scope

Solution Cloud automation/deployment

• Azure Arc, Arc Jumpstart and Agora: Implement end to end automation with Azure Arc being able to deploy complex custom automations easily (Not just infrastructure such as AKS Arc-enabled but the whole solution on multiple sites, orchestrated from Azure cloud). Integration in Azure Arc Agora marketplace.
  • aka.ms/AzureArcJumpstart 
  • https://github.com/microsoft/azure_arc/blob/main/docs/ppt/jumpstart_overview.pptx
  • https://www.youtube.com/@azurearcjumpstart/videos
  • https://azurearcjumpstart.io/

Global aggregated Data in the cloud

• Application data upload to cloud for consolidation and aggregation: Upload events and alerts through Azure Event Hub based on Kafka connector from E4K (Dependency on E4K at the EDGE).
• Global Telemetry uploaded to cloud.
• Storage:: Application data aggregation in Azure CosmosDB and logs into Azure Data Explorer.

Global monitoring, observability, logging and performance/scalability analysis

• Performance analisys, execution time-spans, average per 95% percentile, etc. based on Azure Log Analytics.
• Monitoring/Diagnostics of microservices based on:
  • Azure Monitor for EDGE
  • Application Insights with custom perfkeys.

Application level in the cloud

• Global application alerts dashboard aggregating multiple MEC: "Single Pane of Glass" probably based on "TRIDENT" (aka.PowerBI vNEXT) as a global dashboard aggregating multiple MEC sites (Custom development).
• Global Digital Twin in cloud: Global visibility with Azure Digital Twin about all elements/devices handled in the multiple MEC sites.

Security

• Secrets in Azure KeyVault

CI/CD

• Deployment/CI with GitOps and Flux into AKS EE Arc enabled: Deploy applications to your Arc-enabled AKS Edge Essentials cluster. Example: https://learn.microsoft.com/en-us/azure/aks/hybrid/aks-edge-howto-use-gitops