Why Adopt

Cloudflare Workers run JavaScript at the edge—300+ data centers worldwide. Your code executes milliseconds from your users, not in some distant data center.

FlowLink’s redirect service runs entirely on Workers. The result: P50 latency of 12ms, P99 of 48ms. Try achieving that with a traditional server.

Key Features

1. Global by default - Deploy once, run everywhere
2. Zero cold starts - V8 isolates, not containers
3. KV storage - Globally distributed key-value store
4. Durable Objects - Stateful edge computing
5. Free tier - 100k requests/day

My Architecture Pattern

User Request
    ↓
Cloudflare Worker (nearest edge)
    ↓
KV Lookup (cached at edge)
    ↓
Response (< 50ms)

The Rails app is source of truth for data. Workers handle the read-heavy, latency-sensitive path.

When to Use Workers

Good fit:

• Redirects and proxies
• Authentication/authorization at edge
• API rate limiting
• Image/content transformation
• Geolocation logic

Poor fit:

• CPU-intensive work (50ms CPU limit)
• Long-running tasks
• Complex database queries

The Developer Experience

Workers’ DX is excellent:

• wrangler CLI is fast and intuitive
• Local development via Miniflare
• Git-based deployments
• Excellent observability

Cost Reality

FlowLink handles ~10k redirects/day. Monthly cost: $0 (free tier).

At scale, pricing is $0.50 per million requests. Compare to Lambda’s $0.20/million + duration charges, and Workers often wins on total cost.

For latency-sensitive workloads, Workers is simply the best option.