February 2026 Summaries

In a recent episode of the "Grafana's Big Tent" podcast, experts including Tom Wilkie, Spiros Xanthos, Manoj Acharya, and Cyril Tovena discuss the potential of agentic AI in observability and whether it can autonomously manage production systems. The conversation explores automated root cause analysis, knowledge graphs, and the skepticism faced by site reliability engineers (SREs) towards AI, emphasizing that while AI can assist in debugging and troubleshooting, it is not yet poised to replace human engineers. The panelists highlight the evolution of AI tools in addressing complex production issues, such as a recent incident where AI identified a latent deadlock bug in a system. They also discuss the importance of trust and transparency in AI systems, as well as the challenges related to pricing in an agent-driven world. The conversation concludes with optimism about the future, suggesting that AI's role in observability will grow, potentially achieving autonomy in resolving incidents by the end of the year, akin to the progress seen in AI for coding.

Grafana Labs has introduced a set of Guiding Principles to better reflect its evolved company culture and decision-making processes as it grows. Initially established with aspirational values in 2020, the fast-growing, fully remote company recognized the need to update these to align with its current practices and goals. The Guiding Principles, distinctively tailored to Grafana Labs, emphasize long-term value, integrity, transparency, and customer success, among others, while opposing detrimental behaviors like bureaucracy and mediocrity. These principles are designed to guide daily operations and decision-making, ensuring that the company maintains its unique culture and continues to thrive. The company publicly shares these principles to foster transparency and set clear expectations for potential employees and collaborators.

Grafana 12.4 introduces a suite of enhancements aimed at improving data visualization, observability as code, and user experience. Key updates include Git Sync for managing dashboards as code, Dynamic dashboards for more intuitive navigation, and Suggested dashboards to help users quickly create insights tailored to their data sources. The release also features enhanced visualization options, such as revamped gauge visualizations and time range pan and zoom capabilities, providing users with more flexible and efficient ways to explore data. Additionally, identity and access management are bolstered with SCIM user provisioning and role-based access control for saved queries, while data source integrations see updates with the Zabbix and Google Sheets data sources. The release underscores Grafana's commitment to streamlining workflows and supporting scalable, secure, and customizable data management solutions.

Implementing a robust identity management system in Grafana Cloud is essential for balancing security and scalability as engineering organizations grow. Initially, many teams start with Grafana's default access model, which offers simplicity and rapid productivity but assumes universal account-wide access, posing governance and security challenges over time. To address these, the introduction of a layered identity model using SCIM for automated user and team provisioning is recommended. This model separates administrative control in the Grafana Cloud portal from the engineering workspace within Grafana stacks, allowing for tighter governance at the account level while maintaining a smooth and automated experience for engineers. SCIM integration with identity providers like Okta or Entra ID enables efficient management of access and deprovisioning, reducing exposure from stale accounts and offboarding gaps, and ensuring new engineers have immediate access to necessary resources. This approach mitigates the risks associated with manual access models, which can become operational and security liabilities as organizations scale.

Grafana's latest update introduces the Infinity data source plugin and the Grafana Assistant, enabling users to transform any JSON API into dashboards quickly and efficiently. By leveraging public and private APIs, such as those from USGS, NASA, and CoinGecko, users can access and visualize a wide array of data, including seismic activity, near-Earth objects, and cryptocurrency prices. The Assistant simplifies the process by parsing and transforming data based on user descriptions, allowing for the creation of tailored dashboards with various visualizations like maps and time series. It also facilitates exploratory analysis through direct queries, providing instant answers without the need for full dashboard setups. The Infinity data source supports multiple data formats, though currently, only JSON is compatible with Assistant. Additionally, the tool ensures secure API querying by allowing users to define permitted hosts, maintaining control over data access. This integration is part of Grafana Cloud's offerings, which includes a free tier, making it accessible for developers to enhance their data visualization capabilities.

At Grafana Labs, significant efforts have been made to contribute to the open-source OpenTelemetry project, particularly focusing on the OpenTelemetry .NET SDK, which now supports .NET 10. This achievement was the result of extensive collaboration and testing with the OpenTelemetry .NET instrumentation repositories, ensuring native support for .NET 10, which was released in November 2025. The release cycle of .NET, alternating between Short Term Support (STS) and Long Term Support (LTS) versions since .NET 5, allows the community to test upcoming versions in advance, providing feedback and addressing potential issues, such as the one identified with the .NET logging source generator. The updated OpenTelemetry libraries include new features like schema URL support for metrics and traces, new ASP.NET Core metrics, and a switch to the W3C propagator for the default trace context. This development aims to enhance performance and productivity, offering developers deeper insights and improved monitoring capabilities using OpenTelemetry with .NET 10, alongside observability tools like the Grafana LGTM stack.

In the "Grafana's Big Tent" podcast episode, hosts Mat Ryer and Matt Toback engage with Ted Young, co-founder of OpenTelemetry, to explore the evolution of observability standards and the complexities of tracing adoption. They discuss how OpenTelemetry aims to standardize telemetry data across systems by using a unified language, addressing the historical separation of logging and tracing. The conversation highlights the challenges of implementing distributed tracing due to language-level constraints and the necessity of a consistent context propagation mechanism. Ted Young envisions a future where observability becomes seamless and integrated, with software inherently designed for instrumentation. However, he acknowledges the potential hurdles posed by AI advancements, suggesting that while they may complicate tools, they also hold transformative potential. The episode underscores the importance of maintaining backward compatibility in OpenTelemetry to avoid disruptions, emphasizing the need for careful evolution in a widely adopted open-source project.

Cortex v1.0, a highly scalable and fast Prometheus implementation, has been released for production use, marking a significant milestone for Grafana Labs, which has utilized Cortex in its Grafana Cloud managed logging and metrics platform for several years. As Grafana Labs transitions focus to Grafana Mimir, an open-source long-term storage for Prometheus, Cortex continues to gain traction in large organizations seeking robust, enterprise-ready monitoring solutions. Cortex's new version introduces enhancements such as production documentation, pre-configured Grafana dashboards, and Prometheus alerts to facilitate deployment at scale. The platform supports high availability by deduplicating metrics from HA Prometheus pairs and enables real-time queries against extensive data sets. Widely adopted by companies like Gojek, Cortex has proven its capability to handle large volumes of data, demonstrating its reliability and scalability in diverse production environments. The release has been supported by numerous contributors and remains a valuable resource for those aiming to extend Prometheus's functionality.

Grafana 1.9.0-rc1 has been released, featuring several enhancements and new functionalities without any breaking changes, though as a release candidate, some bugs may exist. Notable updates include the introduction of the Singlestat panel, which offers flexible customization options such as sparklines, color thresholds, and panel links, and enhancements to graph features like a shared tooltip and improved legend table mode. The update also introduces a share panel feature for generating URLs with embedded time ranges and template variables and a new "fill below" style for series visualization. Additionally, panel links can now direct users to other dashboards while preserving time ranges. Torkel Ödegaard, Grafana's creator, plans to attend a metrics meetup in San Francisco hosted by Rackspace on December 3rd, where he will discuss Grafana's new and upcoming features.

GrafanaCON 2026 is set to take place in Barcelona from April 20-22, offering attendees the opportunity to engage with the latest developments in Grafana and its open-source ecosystem. The event promises a range of activities, including hands-on labs, technical deep dives, community-led sessions, and networking opportunities designed to connect Grafana users and enthusiasts. Key highlights include keynote speeches on upcoming releases, opportunities to meet community members in person, and interact with project maintainers. Attendees can also participate in the GrafanaCON Science Fair and the Golden Grot Awards, which celebrate the open-source community's contributions. Early bird registration offers significant discounts, and attendees can choose from various hands-on sessions on topics such as OpenTelemetry instrumentation, AI-assisted plugin development, and advanced dashboarding techniques.

Grafana 8.2.3 has been released to address a medium-severity XSS vulnerability, identified as CVE-2021-41174, affecting all versions from 8.0.0-beta1 to 8.2.2. The vulnerability could allow an attacker to execute arbitrary JavaScript if a user visits a malicious URL while unauthenticated, particularly on pages featuring a login button. Grafana Cloud instances have already been patched, and updated binaries were provided to Grafana Enterprise customers under embargo. Users are advised to upgrade to the patched version or implement workarounds, such as blocking specific URL patterns, to mitigate the issue. An audit has confirmed no exploitation on Grafana Cloud instances, and the issue has been publicly disclosed following a thorough timeline of actions taken since its discovery. Security vulnerabilities should be reported to Grafana via their designated email, and updates are available through their Security Announcements RSS feed.

Grafana Cloud Synthetic Monitoring offers a solution for monitoring internal services within private networks through the use of private probes, which enhance visibility into services like APIs, staging environments, or resources behind firewalls without making them publicly accessible. These private probes are self-hosted agents that run inside a network, performing various checks such as browser, ping, DNS, HTTP, and TCP, and transmit the results back to Grafana Cloud, thus providing observability and alerting capabilities akin to those for public-facing services. The process of setting up a private probe involves deploying a probe agent and configuring synthetic checks within environments like Kubernetes, Docker, or Ubuntu, ensuring that services running on private networks are monitored effectively without exposure to the public internet. This setup allows users to validate service availability and performance from the same environment that internal systems depend on, with the added benefit of Grafana Cloud's generous free tier for getting started with synthetic monitoring.

Grafana Labs is participating in FOSDEM 2024, a major open-source conference held at the Université Libre de Bruxelles, with over 50 Grafanistas attending to engage with the community. The event will feature more than 867 sessions and is expected to draw over 8,000 attendees. Grafana Labs will have a stand showcasing their Grafana LGTM stack and will host the Monitoring and Observability devroom for the eighth year, where topics will include eBPF, CI/CD observability, and distributed tracing. Notable speakers from Grafana Labs, including RichiH Hartmann and others, will present in various sessions across the conference, and attendees are encouraged to visit the Grafana stand or connect with team members throughout the event.

Engineering teams face a recurring decision on whether to build, buy, or leverage open-source solutions for AI-powered observability tools, with each option offering distinct benefits and trade-offs. Building in-house provides maximum control but requires significant time and resources, while buying offers out-of-the-box intelligence that can save time and focus for core competencies. Open source, exemplified by Grafana's ecosystem, offers flexibility and control without starting from scratch, but it necessitates additional responsibilities to create a complete AI solution. Grafana Labs emphasizes that these options are not mutually exclusive and encourages a blend of strategies, allowing teams to start with open-source tools, use managed solutions like Grafana Cloud for acceleration, and integrate custom internal systems as needed. This approach ensures adaptability and continuity in observability strategies, supporting seamless transitions between different lanes on what Grafana describes as a "connected highway" of observability solutions.

Continuous profiling in production environments provides a detailed view of how applications consume resources like CPU and memory, offering insights down to specific lines of code, which helps in identifying root causes of performance issues without needing additional instrumentation or reproducing issues in staging. The practice not only improves system performance by enhancing throughput and reducing latency, but also aids in cutting infrastructure costs and speeding up incident resolution. Despite concerns about overhead, modern sampling profilers, such as those offered by Pyroscope, demonstrate minimal impact on CPU, memory, and latency, typically within low single-digit percentages. These tools use lightweight sampling techniques that avoid extensive instrumentation, making the process efficient and manageable. The backend costs are also optimized using scalable database architectures similar to those in Grafana's systems, ensuring lower storage costs and faster query performance. Teams can measure the actual overhead in their own environments through controlled testing, affirming the cost-effectiveness of continuous profiling as it becomes a standard observability practice.

Grafana has introduced a new feature called playbooks within its Grafana Cloud platform, aiming to enhance the efficiency and consistency of observability practices for organizations with mature systems. These playbooks, now in public preview, allow teams to encode their operational and debugging expertise into structured documents that can be used by their AI-powered Grafana Assistant to provide faster and more consistent responses to system alerts and incidents. Playbooks serve as guides that can include instructions, context, and commands in natural language, and they can be tailored to specific workflows or shared across teams. They offer a dynamic way to manage knowledge and guidelines, allowing for updates post-incident to improve future responses. Playbooks are distinguished from rules, as they are discovered at runtime and offer exhaustive information, while rules remain constant and provide high-level guidance. This feature is designed to be flexible and scalable, encouraging experimentation and iteration to fine-tune processes within organizations.

As Grafana instances grow, managing multiple dashboards through the user interface can become overwhelming, leading to inconsistent tracking and organization. To address this, the concept of managing dashboards as code has been embraced, allowing for better visibility and version control, although it introduces complexity and potential bottlenecks. In response, Grafana has introduced Git Sync, a feature designed to integrate Git version control directly within Grafana, simplifying the process of storing, tracking, and approving dashboard changes. Git Sync allows users to connect Grafana with Git repositories via GitHub Apps or personal tokens, enabling seamless synchronization and pull request workflows directly from Grafana. Currently in public preview, Git Sync aims to expand its capabilities to support more Grafana resources and additional Git platforms like GitLab and BitBucket. Users are encouraged to try Git Sync in development environments, as it is still experimental, and feedback is being gathered to refine the feature further.

Grafana Assistant has introduced playbooks, a new feature designed to integrate AI into existing observability practices, enhancing the efficiency and consistency of incident investigations. Playbooks are structured as simple documents containing titles and instructions, allowing users to encode their organization's operational expertise directly into Grafana Cloud. This tool supports the creation of reusable commands, enabling complex investigation flows to be executed with a single prompt, and can be customized for either individual workflows or broader team applications. Playbooks can also incorporate contextual elements like dashboards and queries, ensuring agents have the necessary resources at their disposal. They differ from rules, which are always applied, by being activated based on user prompts and investigation findings. Best practices include keeping playbooks concise, updating them after major incidents, and experimenting with them before wider deployment. This feature is particularly beneficial for teams with existing complex systems, as it operationalizes their collective SRE knowledge and streamlines investigations.

As the use of agentic AI applications expands, integrating multi-step reasoning, tool execution, and autonomous decision-making into a single workflow, Site Reliability Engineering (SRE) teams require new tools to gain visibility into the performance and behavior of these agents. The OpenAI Agents SDK, a framework for creating agentic applications, includes built-in tracing capabilities to capture detailed agent activity, which can be exported to Grafana Cloud Traces using OpenTelemetry for enhanced analysis. This process involves setting up a sample application, such as a "tutor" app with triage, math, and history agents, and configuring a custom OpenTelemetry tracing processor to send traces to Grafana Cloud. This setup allows developers to visualize and analyze traces for deeper insights into agent workflows and troubleshoot issues efficiently. Additionally, Grafana Cloud Traces offers tools like TraceQL for querying trace data, enabling rich analysis of agent workflows, with the capability to correlate traces with metrics and logs for faster root cause analysis. The article highlights the extensibility of the OpenAI Agents SDK and emphasizes the benefits of using Grafana Cloud for centralized telemetry visualization and analysis, promoting the platform's ease of use and flexible pricing plans.

Prometheus Remote Write v2, introduced in 2024, aims to significantly reduce network egress costs by up to 50% by improving the efficiency of data transmission to Prometheus backends, even in its experimental stage. The new version incorporates first-class support for metadata and implements a symbol table to minimize redundant data, addressing inefficiencies of the original protocol. Grafana Labs has already transitioned to Remote Write v2, achieving substantial cost savings despite a minor increase in CPU and memory usage. This development is especially beneficial for organizations managing large volumes of telemetry, as it enables immediate cost reductions without extensive evaluation. To enable Remote Write v2 in systems like Alloy, users need to configure their setups to accommodate the experimental protocol, with potential for future updates before its final release. The article also hints at further advancements in Prometheus Remote Write, such as replay features and improved resource utilization, which are anticipated to enhance telemetry management further.

Grafana Assistant is a purpose-built AI agent designed to enhance observability by providing transparency and reliability in handling AI hallucinations, which are common challenges in AI models. Unlike typical AI systems, Grafana Assistant leverages live data from Grafana Cloud dashboards, metrics, logs, and traces, ensuring responses are based on actual environmental data rather than generic algorithms. It offers visibility by revealing the queries and data sources behind its answers, allowing users to trace and verify the logic, thus avoiding the "black box" problem. Grafana Assistant also effectively deals with conflicting signals by querying additional data sources and presenting clear patterns or contradictions to guide the next steps, thereby reducing the manual effort in observability tasks. It encourages best practices by turning interactions into shared knowledge, enabling junior engineers to learn while freeing senior engineers from repetitive tasks. This AI assistant acts as a traceable, expert-on-demand partner that aids in maintaining production system stability and enhancing engineering efficiency.