Data Availability Isn’t Observability

The text explores the distinction between data availability and observability in the context of software systems, emphasizing that while high data availability is valuable, observability involves the critical work of extracting meaning from that data. Using a painting analogy, it illustrates how observability helps form mental models by structuring data to reflect user journeys rather than just raw outputs, thus aiding in understanding and debugging. The discussion includes Honeycomb's approach to defining Service Level Objectives (SLOs) aligned with user experiences and highlights the challenges of current practices, such as duplicative alerts and the need for meaningful instrumentation keys. The text advocates for enhancing the interpretability of data by connecting it to product-level concepts, making it more accessible to non-engineers and improving the utility of observability tools across organizations.