Multi-Region Performance Testing

The Challenge

Your agent works great from your office. But what about users in:

Europe (different latency to your servers)
Asia (even higher latency)
Africa (potential routing issues)
Latin America (varying infrastructure)

The Solution

Agent Status tests your agent from real devices in multiple regions, revealing geographic performance differences.

Configuration

Maximum Coverage

{
  "geos": ["us", "eu", "ap", "latam", "af", "me"],
  "max_nodes_per_run": 30,
  "device_types": ["desktop", "mobile"]
}

Focused Testing

{
  "geos": ["us", "eu"],
  "max_nodes_per_run": 10
}

Interpreting Regional Results

Example Results

{
  "overall_verdict": "DEGRADED",
  "by_region": {
    "us": {
      "status": "UP",
      "latency_p50_ms": 450,
      "latency_p95_ms": 890,
      "pass_rate": 100
    },
    "eu": {
      "status": "UP",
      "latency_p50_ms": 680,
      "latency_p95_ms": 1200,
      "pass_rate": 100
    },
    "ap": {
      "status": "DEGRADED",
      "latency_p50_ms": 2400,
      "latency_p95_ms": 4800,
      "pass_rate": 85
    }
  }
}

Analysis

Region	Status	P50 Latency	Issue
US	UP	450ms	Excellent
EU	UP	680ms	Good
AP	DEGRADED	2,400ms	Too slow

Conclusion: Agent servers are in US. EU has acceptable latency. Asia-Pacific users experience poor performance.

Common Patterns

"US Only" Pattern

US: UP, low latency
All others: DEGRADED or DOWN, high latency

Cause: No edge presence outside US. Fix: Deploy CDN or edge nodes.

"Regional Outage" Pattern

Most regions: UP
One region: DOWN

Cause: ISP routing issue, regional DNS failure, or CDN edge problem. Fix: Check regional infrastructure.

"High Latency Everywhere" Pattern

All regions: DEGRADED
Latency: High across the board

Cause: Slow backend (database, upstream APIs). Fix: Optimize server-side performance.

"Intermittent Regional Failures" Pattern

Varying regions fail on different checks
No consistent pattern

Cause: Rate limiting, load balancer issues. Fix: Check capacity and load distribution.

Performance Baselines by Region

Region	Expected P50	Expected P95
US (if servers in US)	200-500ms	500-1000ms
EU	400-800ms	800-1500ms
AP	600-1200ms	1200-2500ms
Other	800-2000ms	2000-4000ms

Optimization Strategies

If Asia-Pacific is slow:

CDN: Cache static assets and API responses at edge

Edge deployment: Deploy inference at edge locations

Regional servers: Spin up servers in AP region

Lighter model: Use smaller model for distant regions

If specific countries fail:

Check DNS: Is DNS resolving correctly in that country?

Check routing: Are there ISP-specific issues?

Check blocking: Is your IP blocked in that country?