Why Your Database Location Matters: The 200ms Lesson

So we deployed Zero Locker and users kept complaining about it being slow. Like, really slow. Everything was taking 2-3 seconds. The weird part? It was instant on localhost.

After blaming the database, the queries, the network, the serverless cold starts - we finally found it. The database was in Frankfurt. The app was in Washington D.C.

What We Actually Saw

• Click on a credential? 2 seconds to load
• Create new credential? Another 2 seconds
• Just navigating around? Everything felt laggy

My first thought was "it works fine locally, so the code is good. Must be something else."

Checked the database - Neon hosted in Frankfurt (eu-central-1).
Checked Vercel - deployed to Washington D.C. (us-east-1).

That's 3,900 miles apart. Every request was crossing the Atlantic Ocean.

The Math

Every request: Washington → Frankfurt → Washington
Round trip time: ~200-300ms
One page load: 5-10 requests
Total delay: 1.5-3 seconds

No wonder it was slow. Each API call was bouncing across continents.

Localhost was fast because everything was on the same machine - network latency was basically zero. Production was slow because physics.

The Fix

We configured Vercel to deploy to Frankfurt:

{
  "functions": {
    "app/api/**/*.ts": {
      "regions": ["fra1"]
    }
  }
}

One config change. That's it.

The Results

Before: 200-300ms per request
After: 5-15ms per request

• API calls dropped from ~760ms to ~120ms
• Page loads went from 2.5 seconds to under 1 second
• The app actually feels fast now

What This Actually Means

The "it works on my machine" thing? For us it wasn't a bug in the code. Everything was fine locally because there was no network latency. In production, every request was adding 200ms just to travel.

A query that takes 10ms to run still adds 200ms for the network. So what should be a 10ms operation becomes 210ms. Do that 8 times on a page load and you're at 1.6 seconds before any processing even happens.

How to Check This

If production feels way slower than localhost, check where everything is deployed:

Find your database region:

• Neon: Check your project dashboard at neon.tech
• Supabase: Check your project settings at supabase.com
• AWS RDS: Check instance details in the console
• PlanetScale: Check region in project settings
• Railway/Render: Check the service details

Find your app region:

• Vercel: Check deployment settings or run vercel inspect
• Netlify: Check site settings
• AWS Lambda: Check function configuration
• Railway/Render: Check service deployment region

Measure the latency: Open browser DevTools → Network tab. Look at API request timings. If you're seeing 100ms+ on every request, you probably have a geography problem.

The fix: Match your app region to your database region. Usually just a config change.

The Actual Impact

After the fix:

• Average response time: 245ms → 28ms
• Bounce rate: 18% → 8%
• Error rate: 2.3% → 0.4%

No code changes. Just moved the functions closer to the database.

Takeaways

1. Match your regions - Database and app should be in the same place
2. Network latency adds up - 200ms × 5 requests = 1 second wasted
3. Localhost lies - It doesn't have this problem, production does
4. Check your metrics - If local is fast but production is slow, it's probably infrastructure

That's the whole story. Geography matters more than you'd think.

Want to check out Zero Locker? Give it a spin - everything runs out of Frankfurt now with sub-20ms response times.