System Design FAQ: Top Questions

A Feature Flag System enables controlled rollout of application features to specific users, teams, or environments. It supports toggling, percentage-based rollouts, A/B testing, and remote control without redeploying code.

📋 Functional Requirements

• Enable/disable flags per user, group, or environment
• Percentage rollout and targeting rules
• API or SDK integration in apps
• Flag evaluation with low latency

📦 Non-Functional Requirements

• Real-time updates (push or polling)
• Audit logging of flag changes
• High availability and fault tolerance

🏗️ Architecture

• Admin UI: For flag creation, rules, and dashboards
• API Gateway: To serve client SDKs
• Evaluation Engine: Applies targeting rules
• Data Store: Redis (fast read) + PostgreSQL (audit/config)

🧪 Flag Rule Evaluation Example

{
  "feature": "new_checkout",
  "targeting": {
    "enabled": true,
    "rollout": 25,
    "rules": [
      {"attribute": "country", "equals": "US"},
      {"attribute": "beta_user", "equals": true}
    ]
  }
}
        

⚙️ Redis Flag Cache Snippet

# Redis Key
GET flags:new_checkout:user123
# Value: true / false / null
        

📡 Client SDK Integration

import { Flags } from 'my-feature-sdk';

const flags = await Flags.getForUser("user123");
if (flags["new_checkout"]) {
  enableNewCheckoutUI();
}
        

🔁 Update Propagation

• Push: WebSocket updates to long-lived SDK connections
• Poll: Periodic fetch every 30 seconds for serverless clients

🔐 Security Considerations

• Flags are signed or obfuscated if exposed client-side
• Audit logs for every toggle/change

📈 Metrics & Audit

• Flag exposure count by version, environment
• Impact on conversion rate or latency (A/B metrics)
• Time to propagate flag change across nodes

🧰 Tools and Tech Stack

• Store: Redis, PostgreSQL
• Frameworks: Unleash, LaunchDarkly (open core)
• UI: React + Tailwind for rule builder

📌 Final Insight

A reliable feature flag system allows safe deployments, real-time experiments, and targeted rollouts. Use Redis for performance, PostgreSQL for durability, and SDKs for consistency across platforms.