Driving down the Cost of Observability Dop224

Title

AWS re:Invent 2023 - Driving down the cost of observability (DOP224)

Summary

  • Chris, a former SRE and Java engineer, now a principal engineer at CoreLogix, discusses strategies to reduce observability costs.
  • Observability costs are often disproportionate to the value provided, sometimes due to inefficiency.
  • CoreLogix offers a SaaS-based observability platform with unique cost optimization tools.
  • The talk covers why observability is expensive, market trends, and how to use data more efficiently.
  • Key points include the high cost of indexing data, the repetitive nature of errors, and the short lifespan of most queries.
  • Chris advises against the common pattern of moving data from high-performance to low-performance storage based on age.
  • He recommends creating usage statistics, defining use cases, and being cautious with indexing.
  • Chris also touches on the importance of avoiding vendor lock-in and asking SaaS providers about their cost optimization tools.
  • CoreLogix's approach to cost optimization includes transparency, use case-driven data management, and leveraging open source tools like OpenTelemetry.

Insights

  • Observability costs can reach up to 30% of a company's cloud budget, which is considered excessive.
  • Microservices, cloud server proliferation, and chaos engineering contribute to increased complexity and data volume, driving up observability costs.
  • A significant portion of observability data is never searched or used, indicating inefficiencies in data management.
  • Chris emphasizes the importance of understanding data usage patterns and aligning them with specific use cases to optimize costs.
  • He criticizes the common practice of tiered data storage based on age and advocates for a use case-driven approach.
  • Open source tools and standards like OpenTelemetry are recommended to avoid vendor lock-in and facilitate easier transitions between providers.
  • CoreLogix's platform focuses on cost transparency and provides tools for customers to manage their data based on usage, which is not common in the industry.
  • The talk suggests that asking SaaS providers about specific cost optimization tools can reveal their commitment to helping customers manage costs effectively.