April 2026 Summaries

kgateway is a Kubernetes Gateway API implementation that utilizes Envoy to manage traffic at the edge of a cluster, making it an essential component for observability in cloud-native architectures. It supports OpenTelemetry for tracing and logging natively via OTLP/gRPC, allowing for seamless integration with observability pipelines, but retains a Prometheus-first approach for metrics collection. This mixed delivery model highlights the complexity of achieving multi-signal observability, where traces and logs are push-based and metrics remain pull-based. By applying a proposed OpenTelemetry Support Maturity Model, which is not an official standard but a community-driven framework, the strengths and gaps in kgateway's OpenTelemetry integration become apparent. The model evaluates dimensions such as integration surface, semantic conventions, and resource identity, aiming to provide a nuanced understanding of where the project excels and where it needs improvement. While kgateway demonstrates robust tracing and logging capabilities, its reliance on Prometheus for metrics and the need for downstream enrichment for resource attributes indicate areas for further development. The analysis underscores the evolving nature of OpenTelemetry support in gateway architectures and suggests that a dimensional maturity model can offer valuable insights into integration challenges and opportunities.

OpenTelemetry has emerged as a crucial component in observability, with nearly half of organizations now utilizing it in production, as noted in the CNCF Annual Survey. It is evolving beyond a mere integration point, becoming the standard mechanism for producing and collecting correlated telemetry across modern platforms. The discussion highlights the importance of OpenTelemetry's predictable behavior at system boundaries, especially at ingress, where requests first enter the cluster. The role of ingress controllers and gateways is crucial as they are the first observable hop for external requests and significantly shape trace data. While tracing has stabilized with most ingress controllers participating cleanly in distributed traces, metrics are still predominantly exposed in Prometheus format, leading to a hybrid model where traces and logs may use OpenTelemetry Protocol (OTLP) while metrics are scraped. The OpenTelemetry Collector plays a significant role in bridging these differences, enriching telemetry, normalizing fields, and ensuring coherence. Semantic precision and resource identity vary across implementations, affecting operational complexity and system trust. As OpenTelemetry becomes foundational infrastructure, the focus shifts from mere support to how well it integrates into a shared telemetry model, with further explorations into specific ingress and gateway implementations planned.

OpenTelemetry has transitioned from an experimental integration to a core component of modern cloud-native infrastructure, as evidenced by its widespread adoption among organizations surveyed by the CNCF. This shift reflects a broader movement towards open standards and vendor-neutral platforms, emphasizing the importance of structured telemetry for comprehensive system observability. Automatic instrumentation provides a quick start, but the real value lies in native instrumentation, which ensures meaningful context and semantics across critical infrastructure components. The challenge lies in maintaining consistent and coherent telemetry, as the traditional "three pillars" of logs, metrics, and traces need to be correlated for effective system understanding. The proposal of an OpenTelemetry Support Maturity Model aims to provide a nuanced framework for evaluating integration quality across multiple dimensions, moving beyond the binary "supports OpenTelemetry" label to better guide architectural decisions and foster trust in shared observability infrastructure. This approach seeks to improve clarity and alignment among project maintainers, platform engineers, and end users, ensuring that telemetry integration evolves in a way that supports robust system-level insight and operational readiness.

OllyGarden and Dash0 have launched Portel, an observability solution specifically designed for portlet-based generative UI architectures, addressing the telemetry quality issues associated with AI-generated portlet fragments. This collaboration stems from the shared experiences of their founders, Juraci Paixão Kröhling and Mirko Novakovic, who both have backgrounds in Java portal servers. Portel is unique in its ability to instrument, trace, and monitor AI-generated portlet fragments across distributed portal containers while adhering to JSR-286 and JSR-362 lifecycle compliance. As enterprises increasingly use generative AI for portal content, traditional observability tools struggle with challenges like portlet-scoped rendering contexts and the non-deterministic nature of LLM-generated markup. OllyGarden's research indicates that AI-generated portlets produce high rates of bad telemetry, which Portel aims to rectify through tools like Rose and Dash0's observability platform. Portel, compatible with major portal server deployments and Kubernetes, offers features like real-time portlet health monitoring and telemetry error detection. The partnership highlights a deepening relationship between OllyGarden, which focuses on improving telemetry quality, and Dash0, an AI-native observability platform, indicating a commitment to addressing an underserved market within enterprise technology.