DataEngBytes 2023 - MEL-T3-02 - Alexei de Lauw
summary
Demonstrates logging and tracing built on Snowflake event table primitives to solve three concrete bottlenecks: debugging long stored procedures and UDFs, avoiding table concurrency locks from many writers, and mitigating micro‑partition fragmentation from single‑row inserts. Implements via UDFs/UDTFs and stored procedures in SQL/Python/Java/JS, uses serverless compute that auto‑scales, shows cost tracking in account_usage.event_usage_history, charges about 1.25 credits for serverless plus cloud services, and enforces spans limited to 128 attributes and 128 events; demo runs and writes real trace and log rows.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:09Underdog features
- 4:12Snowflake logging bottlenecks
- 6:39Logging and tracing solution
- 9:07Logging and tracing setup
- 9:50Logging and tracing costs
- 11:17Demo setup script
- 13:04Python logging procedure
- 13:48Python tracing with spans
- 15:13Querying event logs
- 15:59Viewing tracing data
- 16:40Monitoring Snowflake credits
- 17:05Q&A
related talks
19mDataEngBytes 2023 - SYD-T3-06 - James Weakley
29mDataEngBytes 2023 - SYD-T1-08 - Rishu Saxena & John Cosgrove
17mDataEngBytes 2023 - SYD-T1-05 - Patrick Cuba
26mDataEngBytes 2023 - SYD-T3-01 - Scott Eade
29mDataEngBytes 2023 - BRI-T1-03 - Yousry Mohamed
29mWide Event Analytics
32mDataEngBytes 2023 - BRI-T1-10 Rishu Saxena & John Cosgrove
19mDataEngBytes 2023 - SYD-T1-04 - Yash Sharma
