Performance Bug Roadmap

Do not implement these bottlenecks in the baseline.

Bottleneck	Code change that would introduce it	Expected symptom	Detect with k6	Detect with JMeter	Fix	Metric that should improve
Missing database indexes	Remove foreign key and filter indexes from list/search tables	Slower filtered reads and summary queries	p95 and p99 rise on list/filter scenarios	Longer response times for filtered samplers	Restore indexes on common filters and foreign keys	p95, p99, throughput
N+1 queries	Load relations lazily in list and detail routes	Many repeated queries and slower pages	Dashboard/detail latency climbs under load	Samplers show unstable response times	Use eager loading or fewer round trips	p95, DB query count
No pagination	Return full table data from list endpoints	Larger payloads and slower list pages	List endpoints slow as dataset grows	Read-heavy thread groups degrade quickly	Restore pagination and caps	p90, payload size
Slow search	Use unindexed ilike across many columns	Search endpoints become CPU and DB heavy	Search test p95 increases sharply	Search sampler times spike	Add indexes or full-text search	p95, throughput
Large response payloads	Include nested graphs and verbose fields everywhere	Higher network and JSON parse time	Dashboard/list throughput drops	Response size increases in listeners	Trim response models and add dedicated detail calls	Throughput, response size
Synchronous activity logging	Write logs inline with every request and commit separately	Write latency rises on create/update	CRUD workflow gets slower	Write samplers show longer elapsed times	Buffer or batch log writes	Write latency, throughput
DB lock contention	Update the same rows repeatedly in hot paths	Writes stall under concurrency	CRUD test stalls and error rate rises	Thread group shows blocked samplers	Reduce hot-row updates and lock scope	Error rate, p95
Bad connection pooling	Use tiny pool sizes or no pre-ping	Requests queue up or fail under load	Spike/stress tests show errors	Thread count grows with failures	Tune pool size, pre-ping, timeouts	Error rate, throughput
Inefficient joins	Join too many tables for list endpoints	Higher CPU and slower page loads	Dashboard and list latency increases	Aggregates and view samplers slow	Split queries or precompute summaries	p95, CPU
Expensive summary endpoint	Add many heavy aggregations to /api/stats/summary	Dashboard load becomes a bottleneck	Dashboard smoke/load worsens	Summary sampler dominates report	Cache or precompute summary data	Dashboard p95, DB time