Architectural Comparison

Kolmos vs PostgreSQL:
Row Bloat vs Columnar Economics.

PostgreSQL is the world's gold standard for transactional logic. Kolmos gives you the exact same PostgreSQL wire protocol on Port 5432 while slashing storage bills by 85% to 90%.

Detailed Architectural Matrix

How traditional PostgreSQL compares directly to the Kolmos engine.

Capability
Standard PostgreSQL
Kolmos Cloud Database
Storage ArchitectureRow-oriented heap pages (8KB page size, prone to vacuum bloat)Self-compressing columnar segments (.ksf) on Cloudflare R2
Storage PricingAWS EBS gp3 ($0.08–$0.12/GB/mo) + Provisioned IOPS$0.015/GB/mo on Cloudflare R2 (up to 8× cheaper baseline)
Analytical Query PerformanceRow-by-row CPU execution; CPU bottlenecks on large aggregationsVectorized SIMD execution powered by Apache DataFusion & Arrow
Point Lookup Latency1–2ms via in-memory B-Tree indexes2–4ms via in-memory PK Bloom filter segment sidecars
Compression Ratio1.0× (TOAST compressed strings only)3.5× – 4.0× (Autonomous Formula AST & Prototype medoids)
Application CompatibilityNative PostgreSQL API & Ecosystem100% Native Wire Compatibility (Port 5432 pgwire listener)
Data Egress Fees$0.09 / GB on standard AWS/GCP transfers$0.00 (Zero data egress fees on Cloudflare R2)
Ecosystem ExtensionsExtensive extension library (PostGIS, TimescaleDB, pgvector)Multi-wire protocol engine (Postgres, MySQL, Mongo in one)

When to Choose Standard Postgres

  • Heavy reliance on niche Postgres C-extensions like PostGIS or custom foreign data wrappers (FDWs).
  • Small datasets (< 20 GB) where cloud storage bills are negligible (< $5/month).

When to Choose Kolmos

  • Large datasets (100 GB – 100+ TB) where AWS EBS or Aurora storage bills are exploding.
  • Analytical reporting, dashboards, and aggregations that crawl on standard row-oriented Postgres.
  • Zero code refactoring: connect using existing Prisma, Drizzle, or Django configurations.