Bearuang
A point-of-sale and ERP API verified at 2,000+ requests per minute against 1M+ seeded database rows.
What is this system?
A plain explanation of the project, its scope, and the real tasks it carries out.
Bearuang is a point-of-sale and Enterprise Resource Planning backend for complex retail operations. I built it to stay fast without adding unnecessary complexity.
It has over 30 documented REST endpoints for checkout, catalog and stock management, order fulfillment, and automated reconciliation.
Primary Technologies
Why did it need to exist?
The practical friction, human fatigue, or operational constraints that demanded this solution.
ERP software for small and medium businesses often gets slower as data grows. Heavy ORMs can make queries harder to control, and fragile setup scripts can take hours to configure.
I built Bearuang to keep the software small, its behavior predictable, and deployment quick. A fresh development or staging instance starts in seconds, with complete schema migrations and realistic seed data.
Technical Craft & Architecture
Engineering decisions, query plans, containerization, and measured benchmarks.
Field Benchmark
30+ REST endpoints · 2,000+ req/min (~33 RPS) · 1M+ seeded products
I optimized the SQL queries and containerized the infrastructure to handle concurrent retail workloads without resource spikes.
Load-tested under 2,000+ req/min
Automated load scripts sent ~33 requests per second against PostgreSQL with 1M+ seeded product records. Query latency stayed below a second throughout.
Instant setup via Docker & Kysely
Cut development environment setup from hours to seconds with automated Docker Compose services and type-safe Kysely migrations and seeders.
30+ production REST endpoints
Separated POS transactions, inventory tracking, role-based operations, and automated financial rollups into distinct domains.
Implementation Notes
- Replaced untyped database queries with explicit Kysely query builders for compile-time type safety. The SQL execution plans remain visible, and the queries can use the intended indexes.
- Structured the database schema around composite indexes tailored to frequent POS queries: stock level lookups, product barcode scanning, and multi-tenant ledger entries.
Atmosphere & Feeling
How the project feels in production: its sensory texture and quiet rhythm.
“I picture a sunlit workshop bench with every tool in its place. That’s the feel I wanted from the backend: fast responses, clear HTTP semantics, and queries that don’t keep you waiting.”
A Note from the Field
I pay attention to disk I/O and query execution plans before writing application code. Getting a query over a million rows down to milliseconds means less waiting for the people running the business.
Choose another station stop
Station Stop · STN 03 · Rice Fields (稲田)
Visit another room or return to the main platform.