Feat: Application Deployment Readiness Scoring

[siddhant-khisty]

Description

Often when deploying applications to production environments, DevOps teams need to assess how resilient a workload is for a production environment. This can include things such as its availability, resilience, disaster response, etc. Devtron is looking to implement a mechanism to generate a scorecard that gives a number from 0-100 to denote how ready a workload is for production environments. This scorecard can help a user understand what are the best practices being followed, and what further steps can be taken to improve the production readiness for the workload. Some examples of the criteria include:

• Pod - Topology Spread to ensure spread across multiple AZs
• Autoscaling enabled
  • HPA
  • KEDA
• PDB enabled and disreputable is at least 1
• Quality of Service (QOS)

Deliverable

Analyse a list of set criteria that will analyze the application and generate a scorecard to determine the production readiness for a deployment.

In the scope
Implement a basic scoring mechanism that assesses readiness based on simple criteria such as

• Is a PDB or Topology Spread configured?
• Are the resource quotas well-defined?
• Does the app have liveness & readiness probes?
• Is autoscaling configured? Yes/No
• Separation of workloads across environments such as Dev/UAT/Prod
• High Availability of Nodes - Master & Worker

Future Scope
In the future, we would like to add additional criteria for creating the score card such as

• Security Scanning criteria
• Secret Type - External, CSI
• Autoscaling based on HPA/VPA/KEDA/Karpenter
• State of backup automation
• Multi-region/cluster availability
• Cluster Health
• Monitoring setup

Key Competencies

• Basic Programming
• Curious enough to learn and implement

Recommended Skills to have:

• Go or Python. You can choose the language of your choice
• Kubernetes

Mentors

• Nishant Kumar (@nishant-d)
• Kartik Singhal (@kartik-579)
• Prakhar Katiyar (@prkhrkat)

Skill Level: Easy, Medium
Time: ~175 hrs