Title
AWS re:Invent 2022 - Dow Jones and 3M: Observability with Amazon OpenSearch Service (ANT219)
Summary
- John Handler from AWS introduced the session, emphasizing the importance of search and observability in managing large datasets and identifying critical information.
- Fred Kerwin from 3M Healthcare discussed their observability project, "Astrolabe," which is built on OpenSearch. He highlighted the goals of empowering incident response, enhancing product delivery, and improving interoperability. He also shared their architecture, success metrics, and a success story.
- Krishna Venkatachalam from Dow Jones shared their journey from third-party observability tools to AWS managed OpenSearch Service. He discussed the benefits of migrating to open source tools, their architecture for application logs and tracing, and the success metrics achieved post-transformation.
Insights
- OpenSearch is a community-driven, open-source project that provides search and analytics engine capabilities, which are essential for observability in modern applications.
- The transition from traditional database solutions to search engines like OpenSearch has enabled companies to handle high volumes of log data, build visualizations, and improve real-time monitoring.
- 3M Healthcare's observability project aims to improve incident response, product delivery, and interoperability, leveraging OpenSearch and other AWS services.
- Dow Jones has successfully migrated to AWS managed OpenSearch Service, resulting in cost savings, improved mean time to resolution (MTTR), and better service level objectives (SLOs).
- Both 3M and Dow Jones emphasize the importance of open source solutions for flexibility, cost savings, and avoiding vendor lock-in.
- AWS's commitment to advancing OpenSearch and listening to customer feedback is a key factor in the adoption and improvement of the service.
- The use of AWS managed services and open source tools has enabled both companies to scale their observability solutions and handle larger datasets more efficiently.