Engineering
Performance & Scalability
PaperFox handles thousands of concurrent paper submissions
PaperFox is built to handle large conferences with thousands of concurrent users submitting papers, even during deadline rushes.
Tested Capacity
We regularly load test our infrastructure to ensure reliability during peak submission periods.
1000 Concurrent Users
| Metric | Result |
|---|---|
| Success Rate | 100% (1000/1000) |
| Total Time | ~16.5 seconds |
| PDF Size | 2MB+ |
| Step | Time |
|---|---|
| Authentication | 2-3s |
| File Upload | 2-3s |
| Database Write | 8-11s |
2000 Concurrent Users
| Metric | Result |
|---|---|
| Success Rate | 100% (2000/2000) |
| Total Time | ~30 seconds |
| PDF Size | 2MB+ |
| Step | Time |
|---|---|
| Authentication | 3-4s |
| File Upload | 2-3s |
| Database Write | 22-26s |
5000 Concurrent Users
| Metric | Result |
|---|---|
| Success Rate | 100% (5000/5000) |
| Total Time | ~48 seconds |
| PDF Size | 2MB+ |
| Step | Time |
|---|---|
| Authentication | 1-2s |
| File Upload | 2-16s |
| Database Write | 29-30s |
What This Means
- Large conferences: With 5,000+ concurrent submissions supported, conferences with tens of thousands of submissions can be handled easily since submissions are spread over time
- Deadline rushes: System remains stable during peak submission times
- File uploads: Direct-to-cloud uploads for fast, reliable PDF submission
- No queuing: Every submission processes in real-time, no waiting
Testing Methodology
Our load tests simulate real-world paper submissions:
- User Authentication - Full login flow with session management
- Presigned URL - Secure upload URL generation
- Direct S3 Upload - 2MB+ PDF uploaded directly to cloud storage
- Submission Creation - Complete database transaction with author records
Tests run on Grafana k6 Cloud from distributed AWS load zones, simulating geographically distributed users submitting simultaneously.
Infrastructure
- Database: Neon PostgreSQL with connection pooling
- File Storage: AWS S3 with presigned direct uploads
- Hosting: Vercel Edge Network (global CDN)