Skip to content

docs: Create performance tuning guide for backend under high transaction load #658

Description

@FaveTeamz

Summary

The backend is expected to handle bursts of transaction submissions during popular issue drops. There is no documentation on how to tune the system (connection pool size, Redis cache TTL, Horizon RPC concurrency) for high load.

Problem

Operators facing a traffic spike have no reference guide for which knobs to turn. They must trial-and-error configuration changes during an active incident, increasing MTTR.

Acceptance Criteria

  • Create docs/performance-tuning.md covering:
    • Database connection pool sizing (formula: connections = (core_count * 2) + effective_spindle_count)
    • Redis cache TTL recommendations for global cap vs org assignment queries
    • Horizon RPC concurrency limits and backpressure settings
    • ECS task CPU/memory right-sizing guidelines based on observed benchmark numbers
    • When to scale horizontally vs vertically
  • Include a load testing checklist referencing the k6 test suite
  • Add recommended CloudWatch alarms for saturation signals

Metadata

Metadata

Labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions