Predictive Maintenance with Azure Synapse
summary
A telecom use case processes 10 GB/day of set‑top box telemetry with about 140 event types to replace six‑week Excel reports. The pipeline uses Azure Synapse: serverless SQL on demand, managed Spark notebooks (PySpark), anomaly detection libraries, parquet outputs and Synapse pipelines/spark job definitions to parallelize per‑event modeling. The experiment targets detect anomalies within four hours, validates against historical outages, writes anomaly results to the data lake and surfaces them in embedded Power BI — proven to work end‑to‑end.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 0:41Overview
- 1:21Agenda
- 2:15Quality of Service
- 4:31Anomaly Detection
- 6:47Data Science
- 7:50Process
- 11:29Azure Synapse
- 14:40Azure Portal
- 19:42Synapse Studio
- 20:30Notebooks
- 22:27Spark Pool Definition
- 24:46Performance Optimization
- 26:11Reading and Writing Files
- 27:19Spark Job Definition
- 29:45Recap
- 30:42SQL Scripts
- 34:36PowerBI
- 36:33Azure Data Studio
- 37:24Custom Applications
- 38:20Architectural View
- 39:13Summary
related talks
24mSynapse Analytics
19mEnterprise AI-Powered Predictive Maintenance Platform
51mSupercharge Power BI with Azure Synapse Analytics
30mProcess large-scale data and computational function in serverless manner
50mPower BI + Azure Synapse Analytics: Beginning your journey
46mThe Ultimate Jigsaw - Architecting Azure Synapse Analytics
45mWhat's New and Cool in Azure Synapse Analytics
36mHow to Build a Solution Processing Terabytes of Data (...)?