https://github.com/ltbringer/level-scale
Recipes to scale an application.
https://github.com/ltbringer/level-scale
Last synced: 4 months ago
JSON representation
Recipes to scale an application.
- Host: GitHub
- URL: https://github.com/ltbringer/level-scale
- Owner: ltbringer
- License: mit
- Created: 2025-05-26T00:35:13.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-06-25T23:49:48.000Z (12 months ago)
- Last Synced: 2025-07-19T23:54:05.098Z (11 months ago)
- Language: HCL
- Size: 151 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# 📦 E-commerce System Benchmark: Scaling by Design
This project simulates how a modern **e-commerce backend** might scale from a **solo developer prototype** to a **globally distributed, high-consistency platform**. It's structured around **realistic growth levels**, each introducing new system constraints, architectural shifts, and operational challenges.
You'll work through increasingly complex scenarios covering:
- 🔧 Schema and query design
- 📊 Write-heavy database loads
- 🔁 Replication and partitioning
- 🛠️ Observability and incident response
- 🌐 Global data distribution and fault-tolerance
---
## 🎯 Why E-commerce?
E-commerce systems must prioritize **strong consistency**: orders, inventory, and payment records
**must not be lost or duplicated**. As traffic grows, so do the demands for availability, observability,
and fault isolation. This project reflects real operational trade-offs developers and SREs face daily
when building production-grade systems.
Whether you're tuning WAL settings, scaling read replicas, defining SLAs, or simulating geo-redundant clusters,
this project grows with you -- just like your business would.
---
## 🚀 Level-Based System Scaling Plan (10 RPS → 1M RPS)
| Level | Target RPS | Estimated DAU | Company Stage | Engineering Level | Key Focus Areas |
|-------|------------|---------------|---------------------|------------------------------------|------------------------------------------------------------|
| 1 | 10 RPS | ~1,000 DAU | Solo Dev / MVP | Mid-level engineer | Schema design, single-node PostgreSQL, basic metrics |
| 2 | 100 RPS | ~10,000 DAU | Seed / Pre-Series A | Senior engineer | Indexing, connection pooling, basic joins, dashboards |
| 3 | 1K RPS | ~100,000 DAU | Series A–B | Staff engineer | Partitioning, WAL tuning, write-heavy simulation |
| 4 | 10K RPS | ~1M DAU | Post-Series C | Staff or Principal engineer | Replication, cache, long-join optimization, VPCs, K8s |
| 5 | 100K RPS | ~10M DAU | Pre-global scale | Staff + SRE collaboration | Queues, read/write split, Redis, SLOs, distributed metrics |
| 6 | 1M RPS | ~100M+ DAU | Global distributed | Principal engineer + platform team | Sharding, async writes, geo-routing, chaos testing |
---
## 🧠 Learning Objectives Per Level
### 🟢 Level 1: MVP (10 RPS)
- Functional relational schema
- Basic insert + join operations
- Prometheus + Grafana
### 🔵 Level 2: Small Production (100 RPS)
- Identify slow queries
- Add indexes
- Tune connection pool (e.g., PgBouncer)
- Measure disk vs memory pressure
- SLAs and Incident Management
### 🟡 Level 3: Write Scale (1K RPS)
- Add read replicas
- Add partitions
- WAL tuning
- Richer dashboards
### 🟠 Level 4: Growth Phase (10K RPS)
- Monitor replica lag, cache hit rate
- Compare MongoDB vs Postgres for feed model
- Optimize long joins
- Use query analysis tools
- Custom networking: VPCs
- Kubernetes
### 🔴 Level 5: Pre-Global Scale (100K RPS)
- Introduce query and write split (read/write path separation)
- Redis for caching and rate-limiting
- Optimize background jobs and queues (e.g., like/comment workers)
- Scale monitoring pipelines (Prometheus federation, Grafana Loki)
- External SLA enforcement (e.g., alert budget, SLO tracking)
- Chaos testing
### 🟣 Level 6: Global Distributed (1M RPS)
- Shard database (Citus/Postgres or Mongo)
- Simulate multi-region writes and reads
- Use async models (e.g., event sourcing, log-based ingestion)
- Load test geo-aware routing and failover
- Introduce chaos engineering basics (network partition, node loss)
---
## 🔧 Tooling Overview
| Category | Tool(s) |
|------------------|----------------------------------|
| Metrics | Prometheus, Grafana |
| DB Engines | PostgreSQL, MongoDB (opt-in) |
| Caching/Queue | Redis, Celery/RQ (optional) |
| Load Simulation | Python/Go scripts, ThreadPool |
| Containerization | Docker, Kubernetes, Helm |
| Networking | Custom Docker bridges, mock VPCs |
| Alerts/SLOs | Alertmanager, Loki, Blackbox |
---