Source Repository
This documentation is from amiable-dev/litellm-langfuse-railway.
Last synced: 2026-01-03 | Commit: 5a45454
Starter vs Production: Which Template Do You Need?¶
Quick Decision Guide¶
| You Need... | Choose |
|---|---|
| Quick POC or development | Starter |
| Team evaluation | Starter |
| Side project | Starter |
| Production API | Production |
| Customer-facing product | Production |
| Enterprise deployment | Production |
Feature Comparison¶
Core Features (Both Templates)¶
| Feature | Starter | Production |
|---|---|---|
| LiteLLM Gateway | ✅ | ✅ |
| 100+ LLM Provider Support | ✅ | ✅ |
| Langfuse Observability | ✅ | ✅ |
| Virtual Keys & Budgets | ✅ | ✅ |
| Cost Tracking | ✅ | ✅ |
| Prompt Management | ✅ | ✅ |
| PostgreSQL Database | ✅ | ✅ |
| ClickHouse Analytics | ✅ | ✅ |
| Redis Caching | ✅ | ✅ |
| MinIO Storage | ✅ | ✅ |
Resilience Features¶
| Feature | Starter | Production |
|---|---|---|
| Automated DB Backups | ❌ | ✅ Daily |
| ClickHouse Backups | ❌ | ✅ Daily |
| Backup Retention | ❌ | ✅ 7 days |
| Redis AOF Persistence | ❌ | ✅ |
| Health Monitoring | ❌ | ✅ All services |
| Slack/Discord Alerts | ❌ | ✅ |
| PagerDuty Integration | ❌ | ✅ |
| Prometheus Metrics | ❌ | ✅ |
| Enhanced Restart Policies | Basic | ✅ 10 retries |
| Operations Runbook | ❌ | ✅ |
| Upgrade Documentation | ❌ | ✅ |
Observability¶
| Feature | Starter | Production |
|---|---|---|
| Service Health Dashboard | ❌ | ✅ JSON API |
| Response Time Tracking | ❌ | ✅ Per service |
| Failure Alerting | ❌ | ✅ Configurable |
| Recovery Notifications | ❌ | ✅ |
| Metrics Endpoint | ❌ | ✅ /metrics |
Cost Comparison¶
Starter Template¶
| Service | Est. Monthly |
|---|---|
| LiteLLM | $5-15 |
| Langfuse Web | $5-10 |
| Langfuse Worker | $3-8 |
| PostgreSQL | $5-10 |
| ClickHouse | $5-15 |
| Redis | $3-5 |
| MinIO | $3-5 |
| Total | $29-68/mo |
Production Template¶
| Service | Est. Monthly |
|---|---|
| (All Starter services) | $29-68 |
| Backup Service | $2-5 |
| Health Monitor | $2-5 |
| Total | $33-78/mo |
Production adds ~$4-10/mo for significant resilience improvements.
Use Case Examples¶
Starter is Right For:¶
1. Development Environment - Testing LLM integrations - Prompt development - Team experimentation
2. Early Stage Startup - MVP development - Initial customer pilots - Budget constraints
3. Personal Projects - Side projects - Learning/exploration - Open source contributions
Production is Right For:¶
1. Customer-Facing API - SaaS products using LLMs - AI-powered features - Revenue-generating services
2. Enterprise Internal Tools - Company-wide LLM gateway - Compliance requirements - Audit trails needed
3. High-Stakes Applications - Healthcare AI assistants - Financial analysis tools - Legal document processing
Migration Path¶
Starter → Production¶
If you start with Starter and need to upgrade:
- Deploy Production template alongside existing
- Migrate data using pg_dump/restore
- Update DNS/endpoints to point to new stack
- Verify functionality
- Decommission Starter stack
Timeline: 1-2 hours for small deployments¶
FAQ¶
Q: Can I add backups to Starter myself?
A: Yes, but Production template saves you time: - Pre-configured backup service - Tested retention policies - Alert integration included - Documentation for restore
Q: Is Production overkill for my use case?
A: Ask yourself: - Would 4 hours of downtime cost you customers? - Would data loss set you back weeks? - Do you need audit trails?
If any answer is "yes," Production is worth $5-10/mo extra.
Q: Can I upgrade later?
A: Yes, but it's easier to start with Production if you expect to need it. Migration requires: - Database export/import - Service reconfiguration - Testing period
Q: What about Railway's HA PostgreSQL?
A: Production template is designed to work with it. See UPGRADE.md for migration steps when you're ready for even higher availability.
Summary¶
| Criteria | Starter | Production |
|---|---|---|
| Setup Time | 5 min | 5 min |
| Monthly Cost | $30-70 | $35-80 |
| Recovery Time | Hours-Days | Minutes |
| Data Loss Risk | High | Low |
| Alert Coverage | None | Full |
| Documentation | Basic | Comprehensive |
| Maintenance Burden | Higher | Lower |
Our Recommendation: - Experimenting? → Starter - Building a product? → Production from Day 1