Idempotent Writes, and Why Retries Duplicate Data
Every pipeline gets retried eventually, and a non-idempotent write turns each retry into duplicates. Four patterns that make a write safe to repeat.
10 articles
Every pipeline gets retried eventually, and a non-idempotent write turns each retry into duplicates. Four patterns that make a write safe to repeat.
Configure pipelines to run on schedules or in response to data changes, enabling fully automated ML operations with minimal manual intervention.
Data Pipelines in Fabric are similar to Azure Data Factory pipelines but with deeper Fabric integration.
Checks provide automated validation gates for enterprise deployments.
Azure Pipelines templates and stages enable maintainable, scalable CI/CD implementations.
Azure ML Pipelines v2 provides a modern, Pythonic way to build production-ready ML workflows.
Going into 2022, the Azure Pipelines features worth adopting are the security controls: environments, checks, protected repos and scoped tokens.
The first ML pipeline I helped build was a series of Python scripts duct-taped together with a bash wrapper and a nightly cron job. It worked exactly as…
The first ML model I helped put into production was a notebook a data scientist ran by hand every Monday morning. That worked exactly as well as you'd…
Templates transform pipeline chaos into maintainable, standardized CI/CD.