Completek8s Classroom notes 02/Nov/2023

Why Kubernetes is Challenging in Production

• k8s in production brings challenges and difficulties from
  • scaling
  • uptime
  • security
  • observability
  • resource utilization
  • cost management
• K8s lacks complete support for some essential services such as IAM, storage and image registries
• Learning curve and too many moving parts make it little bit more difficult to managing k8s. Lets have a look at k8s infra layers
  

Production Readiness checklist

Cluster Infrastructre

• The following check list items cover the production readiness requirements on the cluster level
  • Run a highly available control plane
  • Run a highly available workers group
  • Use a shared storage management system
  • Deploy infrastructure oservability stack

Cluster services

• The following checklist items cover the production readiness requirements on the cluster services level
  • Control cluster access
  • Hardening the default pod security admission
  • Enforce custom policies and rules
  • Deploy and restrict network policies
  • Enforce Security checks and conformance testing
  • Deploy a backup and restore solution
  • Deploy an observability stack for cluster componets

Apps and Deployments

• The following checklist items cover the production readiness requirements on the apps and deployments level
  • Automate images quality and vulnerability scanning
  • Deploy ingress controller
  • Manage certificates and secrets
  • Deploy app observability stack

Kubernetes Infrastructure Best Practices

The 12 principles of infrastructure design and management

• The following list summarizes the core principles that may lead to decision making through the k8s infrastructre desing process.
  • Go Managed
  • Simplify
  • Everything as Code (Xac)
  • Immutable infrastructure
  • Automation (GitOps and Operators)
  • Standardization
  • Single Source of truth (Git)
  • Design for availability
  • Cloud agnostic
  • Business Continuity
  • Plan for failures
  • Operational effeciency

Cloud Native landscape & ecosystem

• This landscape has four layers
  • Provisioning
  • Runtime
  • Orchestration Managment
  • App definition and development
• Cloud Native Trail map Refer Here

Best-Practices for production

• List of important considerations and best practices to run k8s in production
  • Cluster Configuration:
    • Use infrastructure as code (IaC) to automate the creation and management of k8s clusters
    • Seperate your clusters for development, testing and production
  • Security
    • Following the priniciple of lest privilege to access the k8s api
    • Use RBAC to manage access to resources
    • Secure your cluster with network policiews
    • Use naamespaces to isolate workloads
    • Keep container images free of vulnerabilities and regularly scan them
    • Use trusted base images for containers
    • Enable audit loggging to keep track of activities
  • Networking
    • Use CNI with network policies enabled
    • Expose services through ingress controllers and LoadBalancer with secure connectivity
  • Storage:
    • Use persitent volumes for stateful applications
    • Regularly backup your persisted data
    • Implement robust storage solutions that match your IOPS and throughput requirements
  • Monitoring & logging:
    • Implement a comprehensive monitoring solution like Prometheus to track cluster state and Performance
    • Aggregate and analyze logs using tools like Elastic search, fluentd and Kibana
  • High Availability:
    • Run k8s control plane components in HA mode
    • Deploy critical applications with multiple replicase
    • Distribute workloads across multiple nodes and zones
  • Disaster Recovery:
    • Create a regular backups of your cluster state (etcd)
    • Have a Disaster Recovery Plan in place
  • Automation:
    • Automate your deployments with CI/CD pipelines
    • Use GitOps for declarative infrastructure and applications management
  • Resource Management:
    • Implement Resource requests and limits to ensure fair scheduling and avoid resource contention
    • Use Horizontal Pod Autoscaling to adjust number of pod replicase based on Lod
  • Updates and Upgrades:
    • Regularly apply updates to k8s and containerized applications
    • Perform rolling updates to minimize downtime
  • Performance Tuning:
    • Profile the performace of your applications and optimize them as needed
    • Tune the kernel and network settings for better performance where necessary
  • State Managment:
    • Use stateful sets for workloads that require stable and persitent storage
    • Ensure the state is backed up to avoid the data inconsistency
  • Cost Management:
    • Monitor resource usage to optimize costs
    • use cost-allocation tags for billing and cost optimizations