Observability and SRE
摘要
Operating Snowflake well means turning platform behavior into queries you can run every day. Instead of log files and agents, you get first-class system views that describe queries, warehouses, costs, storage, security events, replication, and data movement. The SRE posture is simple: collect a small set of golden queries, materialize them into an ops schema, and wire alerts from SQL, not shell scripts. When something goes wrong, investigations start with the same views you already use for weekly health reports.