Back to Engineering Blog
Architecture8 min read

Why We Chose Single-Table Design for Our Database

Deep dive into composite keys, index partition strategies, and handling deployment records at scale.

Sarah Chen
Sarah Chen
Principal Database Architect • Published August 2, 2026

Single-table design in a key-value database is one of the most effective techniques for building hyper-scalable SaaS backends. By modeling multiple entity types in a single table using composite primary keys (PK and SK), we achieve O(1) query performance across all application entities.

Key Schema Design for ServerLaas

Our core database schema stores Users, Projects, Deployments, and Billing Subscriptions inside a single table named 'serverlaas-main':

json
{
  "PK": "USER#usr_99812",
  "SK": "METADATA",
  "email": "dev@obsidianx.online",
  "plan": "PRO"
}

Global Secondary Index (GSI) Partitioning

To query deployments by status (e.g. FETCH_ALL_BUILDING_PROJECTS), we leverage GSI1 with GSI1PK set to STATUS#BUILDING and GSI1SK set to the ISO timestamp.

Related Articles