Open Observability Summit + OTel Community Day 2025

Hybrid cloud isn’t a stepping stone—it’s a destination. With 39% of CNCF survey respondents already operating in hybrid environments, this model is here to stay. But as teams pursue cloud-native architectures, many skip a critical step: developing a clear cloud strategy and an observability approach to match.
The result is predictable— widening visibility gaps, redundant tooling and data, and spiraling costs as teams try to stitch together disconnected, vendor-specific systems never meant to work in concert. Hybrid environments expose these issues quickly, especially when workloads span multiple platforms without a unified way to observe and understand them.
Modernization efforts demand open observability from the start—not as an add-on. Technologies like OpenTelemetry, Fluent Bit, and Prometheus act as connective tissue across clouds, clusters, and on-prem infrastructure, enabling standardization where it’s needed most.
This talk outlines how to center open observability in your modernization journey: where to standardize architectural layers, how to maintain a more open approach, and why these decisions have long-term payoff. 
Hybrid complexity is inevitable. Leading with open observability is how you stay in control—now and in the future.

Logs can get very expensive and often how useful all those logs are is unknown, some are but many are not. It is very difficult to know which logs are useful and how exactly they are used. With Chronosphere's Control plane for logs users can now get a comprehensive analysis of value and usage patterns, along with sophisticated recommendations and control actions that allow some or most of the value derived from those logs to be preserved. In order to achieve our goals we have enhanced Fluent Bit to be more flexible in which logs are actioned upon and will share useful future additions to it.

OpenTelemetry has accelerated the commoditization of instrumentation. Telemetry generation is becoming a solved problem, an implementation detail. But this has created a new challenge: a wealth of standardized signals with no standard meaning. Different systems instrumented with different semantics generating telemetry in their own unique language. And while signal correlation connects specific workloads, it fails when we need to understand our systems at a macro scale by joining disparate datasets.
That is, until we all agreed to speak the same language.

Just as English as a lingua franca fueled progress across the internet, OpenTelemetry Semantic Conventions are providing a shared language for our systems. In this talk we’ll discuss why semantic interoperability is the real connective tissue, how it’s fueling deeper insights into our production environments, and the key role it plays in enabling the AI systems that are rapidly ushering in the next revolution of our industry.

Today, we face distributed systems of exascale complexity that far exceed human comprehension. Traditional observability tools are no longer enough. But what if we could fundamentally redefine observability?

Enter Large Language Models. LLMs aren't just tools anymore; they're becoming the entire observability toolchain.

This talk will explore compelling, real-world examples demonstrating how LLMs are redefining what's possible. The observability paradigm shift is here, enabling profound understanding, rapid prediction of actionable insights, and the realization of truly self-healing systems. Embrace the future. It's already here.