Databricks 可观察性：使用系统表监视和管理 Databricks 实例.pdf

1、Databricks ObservabilityUsing System Tables to Monitor and Manage Your Databricks InstanceGreg KroleskiMichael PostiglioneAgenda3What is observability?Michael PostiglioneHow does Databricks enable observability?Greg KroleskiData CollectionGreg Kroleski&Michael PostiglioneDatabricks Observability Too

2、lsMichael PostiglioneWrap Up+Q&AGreg Kroleski2025 Databricks Inc.-All rights reservedWhat is observability?2025 Databricks Inc.-All rights reservedObservability is the way you understand a system so you anticipate and prevent problems2025 Databricks Inc.-All rights reservedObservability Explained En

3、suring your data intelligence platform is working as expected requires:62025 Databricks Inc.-All rights reservedData CollectionMonitoringDetectionRemediationAnalysisGather,aggregate,and structure telemetry data from across your platformContinuously track system health and behavior using telemetry da

4、taIdentify anomalies,failures,or performance regressionsInvestigate signals to determine the issue root cause and scopeImplement targeted fixes or adjustments to restore expected system behaviorObservability Personas 7FinOpsDescription:Tracking and optimizing cloud usage and spend across services,us

5、ers,and workloadsSecOpsDescription:Observing and correlating security events and data access to detect threats and ensure complianceDevOps-InfraDescription:Monitoring and analyzing E2E infrastructure and application performanceDevOps-DataDescription:Maintaining the health,reliability,and freshness o

6、f data pipelines and datasetsDevOps-AI&MLDescription:Tracking and explaining machine learning model performance,drift,and fairness in production2025 Databricks Inc.-All rights reservedObservability CUJs 8FinOpsHow much are we spending?Which projects are costing the most?How much should we budget for