David Henner - Level up performance with simple coding changes - Rails World 2024

Add proper observability and monitoring to identify performance bottlenecks - tools like Datadog traces help pinpoint issues

Leverage proper caching strategies:

• Use read_multi instead of multiple individual cache reads
• Consider ActiveSupport memory store for frequently accessed data
• Implement ETags and 304 responses to reduce unnecessary data transfer

Optimize database queries:

• Use pluck instead of instantiating full ActiveRecord objects when possible
• Replace array operations with more efficient data structures (Set vs Array.unique)
• Use hashes for lookups instead of array iterations
• Be careful with forced indexes as they may become problematic over time

Implement proper data lifecycle management:

• Plan for data archiving from the start
• Consider sharding for large datasets
• Have a clear data retention/removal strategy

Write performance-focused tests:

• Test for specific numbers of SQL queries
• Add assertions for memoization behavior
• Verify cache hits/misses

Optimize code patterns:

• Use flat_map instead of map.flatten
• Properly memoize shared data on singleton/global objects
• Process/sanitize data on input rather than output
• Use blocks for custom performance tracing

Small optimizations add up:

• Even 1ms improvements matter at scale (billions of requests)
• Focus on high-traffic endpoints first
• Consider infrastructure costs alongside latency