profiling-transaction-fingerprints

What problem does it solve?

Analyzing historical transaction performance to identify high-retry transactions, contention patterns, and commit latency issues using aggregated and sampled statistics from crdb_internal.transaction_statistics, enabling root-cause analysis without DB Console access.

Core Features & Use Cases

• Transaction-level profiling to surface bursts of retries and latency, helping identify hot rows and contention hotspots.
• Statement-level drill-down by linking transaction fingerprints to statement statistics to reveal problematic queries within high-retry transactions.
• Resource and performance diagnostics by correlating retry/commit latency with network, memory, and disk usage metrics for optimization.

Quick Start

Execute a set of queries to surface top transaction fingerprints from the last 24 hours, then inspect their statement composition to pinpoint bottlenecks.