Reduce RDS Cost 50%: Custom Cache Layer Case Study

A catalogue and session-heavy product was spending a large share of its AWS bill on RDS. Read replicas and bigger instances had not solved latency; they had increased cost.

Designed and built a custom in-memory layer (VeloxDB) for hot catalogue data and session state. The layer ran on two EC2 instances; the application read from it for the hot path and wrote through to RDS for durability. RDS could be downsized (fewer replicas, smaller instance type).

RDS spend dropped by over 50%. P99 latency on the hot path dropped by an order of magnitude. The custom layer paid for itself in under a year from infra savings alone.

When the hot path is a well-defined subset of your data, a purpose-built cache is often cheaper and faster than scaling a general-purpose database. At Hendoi we build custom database engines for USA, Canada, and Bengaluru. Contact us for a free cost assessment.

📞 +91-9677261485 | 📧 support@hendoi.in | Contact us