All Products
Search

This Product

    Cloud Monitor:Enable observability on ACK with OBI

    Document Center

    Cloud Monitor:Enable observability on ACK with OBI

    Last Updated:Jun 11, 2026

    OpenTelemetry eBPF Instrumentation (OBI) provides non-intrusive monitoring to report trace data to CloudMonitor 2.0. CloudMonitor 2.0 then starts monitoring your applications, allowing you to view data such as application topology, traces, exception transactions, slow transactions, and SQL analysis.

    OBI overview

    OBI is built on eBPF and deployed as a DaemonSet on cluster nodes. It automatically collects application metrics and traces without requiring you to modify application code or inject a sidecar.

    OBI automatically discovers the following protocols:

    Protocol

    Description

    HTTP

    Traces HTTP/1.x and HTTP/2 requests and responses.

    gRPC

    Traces gRPC unary and streaming calls.

    SQL

    Traces queries for databases such as MySQL and PostgreSQL.

    Redis

    Traces Redis command calls.

    Kafka

    Traces Kafka producer and consumer messages.

    GenAI

    Traces AI calls, including OpenAI, Anthropic, Tongyi Qianwen, MCP, Embedding, and Rerank.

    OBI exports the collected telemetry data to Managed Service for OpenTelemetry over OTLP (OpenTelemetry Protocol).

    Install and configure OBI

    Install OBI

    1. Log on to the Container Service for Kubernetes console. In the left-side navigation pane, click Cluster List, and then click the name of the target cluster.

    2. In the left-side navigation pane, click Component Management. Find the arms-obi component and click Install.

    3. Click Next to complete the installation. Adjust the CPU and memory limits as needed.

      Note

      If a node runs many applications, consider increasing the CPU and memory limits (MEM Limit) to prevent data collection failures caused by insufficient resources.

      Scenario

      CPU request / limit

      Memory request / limit

      Light load (4 pods, total QPS < 1,000)

      100m / 250m

      128Mi / 256Mi

      Medium load (4–8 pods, total QPS 1,000–5,000)

      250m / 500m

      256Mi / 512Mi

      High load (8 pods, total QPS > 5,000)

      500m / 1000m

      512Mi / 1Gi

      Heavy load (8+ pods, total QPS > 20,000)

      1000m / 2000m

      1Gi / 2Gi

    Configure monitoring targets

    Note

    Restart applications that use the gRPC protocol to enable monitoring.

    1. In the Container Service for Kubernetes console, navigate to Configurations > ConfigMaps. Switch the namespace to obi-system, find arms-obi-discovery-default, and then click Edit.

    2. Add the k8s_deployment_name, k8s_namespace, and the name for the application that will be displayed in CloudMonitor 2.0. Example:

      discovery：
  instrument：
    - k8s_deployment_name: "server"
      k8s_namespace: "arms-default"
      name: "server-test-obi"

    Verify the monitoring data

    After you configure OBI, wait about two minutes for the monitoring data to appear.

    1. Log on to the CloudMonitor 2.0 console. Select the target workspace. In the left-side navigation pane, go to All Functions > Application Observability > Application Monitoring.

    2. On the Application List page, click the name of the target application to view monitoring data such as application topology, traces, exception transactions, slow transactions, and SQL analysis. For more information, see Application Monitoring.