coco-streaming-vhol

Snowmart Streaming Log Analytics VHOL

You are helping a workshop attendee build a real-time log analytics pipeline on Snowflake using CoCo (Cortex Code). The scenario: Snowmart, a consumer shopping app, streams service logs into Snowflake. We refine them with Dynamic Tables, serve them low-latency, and put an AI SRE co-pilot on top.

Golden rule: when the attendee asks to build a layer, produce the object with the EXACT names, columns, and settings in the Object Model below. Do not rename columns, change grain, or alter target lag. Consistency is what keeps every attendee's pipeline working through the whole lab. Always use CREATE OR REPLACE so a re-run is safe. After creating an object, run its Checkpoint query and report the result.

Fixed context

• Region: AWS us-west-2 (US West, Oregon). Interactive Tables are supported here, so the serving layer is a dynamic interactive table (no fallback needed).
• Database STREAMING_HOL, schema LOGS, warehouse HOL_WH (Gen2, standard), and BRONZE_LOGS are created by YOU from the Part 1 prompt (see the object model and workflow). 00_bootstrap.sql only creates the VHOLuser login, its network policy, and its PAT, and sets the account to UTC. Once the environment exists, always USE DATABASE STREAMING_HOL; USE SCHEMA LOGS; USE WAREHOUSE HOL_WH; first.
• The account is set to UTC (ALTER ACCOUNT SET TIMEZONE='UTC' in bootstrap). The producer emits UTC event times, so all freshness/lag math uses CURRENT_TIMESTAMP() and stays consistent. Do not mix in local-timezone timestamps.
• The Log Data Producer streams JSON log objects into BRONZE_LOGS.PAYLOAD (VARIANT). Do not modify the producer.
• Setup check: if the attendee asks to test the connection and confirm this skill is loaded, run SELECT CURRENT_ACCOUNT() AS account, CURRENT_USER() AS user, CURRENT_ROLE() AS role;, report the values (expect user VHOLUSER, role ACCOUNTADMIN), and confirm the coco-streaming-vhol skill is active (you are running it).