All engineering cases
CASE 02 / Performance · E-commerce

Investigating database timeouts under e-commerce traffic.

After an AWS-to-GCP migration, a high-traffic e-commerce backend started timing out against the database under load. I tested pooling, caching, cold starts and autoscaling before proposing changes.

Context

I worked in a high-traffic e-commerce environment after its infrastructure migration from AWS to GCP.

Problem

The system began experiencing database timeouts under higher traffic. It was not obvious which layer was failing: the database, the application, or the new cloud setup.

Investigation

  1. Used load testing and concurrency analysis to investigate behavior under traffic.
  2. Examined cold starts and database connection pooling.
  3. Analyzed caching and autoscaling alongside database and infrastructure bottlenecks.

Root cause

The timeouts were not one misconfigured parameter. They came from the interaction of cold starts, connection pooling, caching and autoscaling after the migration.

My contribution

  • Investigated system behavior under load instead of guessing at a fix.
  • Tested hypotheses before proposing architecture changes.
  • Analyzed database and infrastructure bottlenecks.
  • Contributed to improvements involving pooling, caching and autoscaling.
THE RESULT

Outcome

Pooling, caching and autoscaling changes followed the investigation, and the system handled traffic more reliably afterwards.

What I learned

Performance problems are rarely solved by changing one parameter. Measure each layer before deciding where to optimize.
NEXT CASE / 03Standing up a shared React system for a US healthcare team
Discuss my experience