Title

AWS re:Invent 2022 - How to integrate analytics into DevOps processes using Git (PRT098)

Summary

• The speaker, a former software developer, emphasizes the importance of analytics teams adopting software engineering methodologies, particularly Source Control Management (SCM), to improve their development processes.
• The speaker highlights the common practice of pushing analytics changes directly to production without proper testing, staging, or committing, which can lead to DevOps breakdowns and conflicts.
• The talk discusses the challenges of integrating analytics tools, which often lack SCM features, into a DevOps workflow, and how most analytics platforms are not built with Git support, making it difficult to track changes and manage deployments.
• The speaker outlines critical capabilities for implementing DevOps with embedded analytics platforms, including expressing everything in source code format, using the same branching strategy as the rest of the product, ensuring easy switching between branches, and having a proper Git interface.
• Sisense, the speaker's company, is presented as a solution that natively supports Git, allowing for seamless integration of analytics into existing DevOps processes, and enabling development, testing, and release with confidence.

Insights

• The integration of analytics into product offerings is becoming increasingly common, and there is a growing need for analytics platforms to support DevOps practices.
• Traditional BI tools often lack the necessary features to integrate smoothly with SCM systems like Git, which can lead to inefficient and risky development practices.
• The speaker's experience with Sisense suggests that it is possible for analytics platforms to fully support Git, which can greatly enhance the development workflow for analytics within a product.
• The ability to express analytics components (charts, dashboards, models) in source code format (e.g., JSON or YAML) is crucial for effective version control and CI/CD practices.
• The talk underscores the importance of analytics platforms offering the flexibility to use standard software development practices, such as feature branching and continuous delivery models, to ensure that analytics can be developed, tested, and released with the same level of confidence as other software components.
• The speaker's call to action for analytics vendors is to provide full support for SCM practices, which would benefit both software and analytics teams and lead to more robust and reliable analytics integrations in products.