sentry-performance-monitoring

Transactions track high-level user flows, while spans measure sub-operations within them. Initialize Sentry, wrap your code in a transaction, create child spans for specific operations, and capture timing and metadata automatically or manually to visualize performance bottlenecks across your application.

Yes, distributed tracing propagates trace context across services. Sentry automatically captures and links transactions and spans from frontend and backend codebases, allowing you to follow a single user request through your entire system and identify where latency occurs.

Initialize Sentry with your DSN, enable browser tracing integration if monitoring frontend code, configure sampling and profiling settings to control data volume, then wrap critical user flows in transactions. Sentry automatically instruments many operations; add manual spans for custom logic requiring measurement.

Transactions represent complete user operations or requests; spans are nested measurements of individual tasks within a transaction. A payment flow is a transaction; database queries, API calls, and cache lookups are child spans that break down where time is spent.

No, Sentry automatically instruments many frameworks and libraries. Start with transactions around critical user flows, then add manual spans only where automatic instrumentation gaps exist or where custom business logic requires visibility into performance characteristics.

Configure sampling rules to capture a percentage of transactions or filter by criteria like endpoint or user. Profiling settings let you enable or disable CPU and memory profiling separately, and you can set custom sampling functions to balance observability with data costs.