Attribute Sensor is Now Officially Certified for GKE Autopilot

In the ever-evolving landscape of cloud-native technologies, Google Kubernetes Engine (GKE) Autopilot is simplifying Kubernetes operations by managing the underlying infrastructure. We are thrilled to announce that the Attribute Sensor is now officially certified for GKE Autopilot, empowering GCP customers with deep, real-time cost and resource utilization visibility for Autopilot workloads.

This certification means Google Cloud customers using Autopilot, can now easily adopt Attribute’s FinOps zero tagging technology, and access  granular application-layer costs over teams, products and customers. 

The Power of eBPF on GKE Autopilot

GKE Autopilot streamlines cluster management by abstracting away node configuration and management. This managed environment, however, presents unique considerations for monitoring and security tools that require deeper access to the underlying kernel.

This is where the power of eBPF (extended Berkeley Packet Filter) comes in. The Attribute Sensor is built on eBPF technology, which allows it to gain deep visibility into resource consumption at the application level through packet inspection. This technology operates within a secure, sandboxed environment in the kernel with read-only permissions, ensuring both advanced security and low operational risk.

Our support for GKE Autopilot is made possible through Google’s partner program, which allows select, verified solutions to run with the necessary privileges in Autopilot clusters. This ensures that you can benefit from the Attribute Sensor’s capabilities without compromising the security and managed nature of your Autopilot environment.

Why Attribute Sensor for Your GKE Autopilot Clusters?

As you embrace the operational efficiencies of GKE Autopilot, understanding your cost drivers becomes even more critical. The Attribute Sensor provides the missing piece of the puzzle by:

• Holistic Cost Understanding: Move beyond basic infrastructure costs and attribute expenses to specific features, applications, teams, and customers.
• Granular Resource Utilization: By inspecting network packets, the sensor generates and aggregates usage metrics across all your workloads, including shared resources. For instance, it can pinpoint the resources consumed by a specific microservice’s API call.
• Optimized for Performance: The sensor is designed for high-traffic environments, aggregating usage metrics within the cluster before transmitting data to the Attribute backend. This minimizes data transfer costs and is optimized for large-scale deployments. Performance tests have shown the sensor has a negligible impact on application performance, with CPU usage remaining within a 5% cap even under peak loads.
• Security-First Design: The Attribute Sensor operates with a security-first approach. All data is transferred over TLS encrypted connections, and any sensitive information is stripped within your environment before transmission. The eBPF program itself undergoes strict verification to prevent any potential exploits.

Getting Started with Attribute on GKE Autopilot

Deploying the Attribute Sensor in your GKE Autopilot environment is a straightforward process. The sensor is deployed using an operator, and our dynamic configuration support extends to environments leveraging GCP Autopilot.

To learn more and get specific installation documentation for your environment, please contact our support team at [email protected].