Amazon Bedrock → OpenObserve

Automatically capture token usage, latency, and model metadata for every Amazon Bedrock converse call in your Python application.

Prerequisites

Python 3.8+
An OpenObserve account (cloud or self-hosted)
Your OpenObserve organisation ID and Base64-encoded auth token
AWS credentials with AmazonBedrockFullAccess permissions

Installation

pip install openobserve-telemetry-sdk openinference-instrumentation-bedrock boto3 python-dotenv

Configuration

Create a .env file in your project root:

OPENOBSERVE_URL=https://api.openobserve.ai/
OPENOBSERVE_ORG=your_org_id
OPENOBSERVE_AUTH_TOKEN=Basic <your_base64_token>

AWS_ACCESS_KEY_ID=your-access-key-id
AWS_SECRET_ACCESS_KEY=your-secret-access-key
AWS_DEFAULT_REGION=your-aws-region

Instrumentation

Call BedrockInstrumentor().instrument() before creating any boto3 client.

from dotenv import load_dotenv
load_dotenv()

from openinference.instrumentation.bedrock import BedrockInstrumentor
from openobserve import openobserve_init

BedrockInstrumentor().instrument()
openobserve_init(resource_attributes={"service.name": "amazon-bedrock"})

import os
import boto3

bedrock = boto3.client(
    "bedrock-runtime",
    region_name=os.environ.get("AWS_DEFAULT_REGION", "us-east-1"),
)

response = bedrock.converse(
    modelId="amazon.nova-lite-v1:0",
    messages=[{"role": "user", "content": [{"text": "Explain observability in one sentence."}]}],
)
print(response["output"]["message"]["content"][0]["text"])

What Gets Captured

Attribute	Description
`operation_name`	Always `bedrock.converse`
`llm_model_name`	Model ID used (e.g. `amazon.nova-lite-v1:0`)
`llm_token_count_prompt`	Input tokens consumed
`llm_token_count_completion`	Output tokens generated
`llm_token_count_total`	Total tokens for the request
`llm_input_messages_0_message_role`	Role of the input message (`user`)
`llm_input_messages_0_message_contents_0_message_content_text`	Input message text
`llm_output_messages_0_message_role`	Role of the output message (`assistant`)
`llm_output_messages_0_message_content`	Full response text
`input_value`	Raw prompt text
`output_value`	Raw response text
`openinference_span_kind`	Always `LLM`
`span_status`	`UNSET` on success, `ERROR` on failure
`duration`	End-to-end request latency

Viewing Traces

Log in to OpenObserve and navigate to Traces in the left sidebar
Click any bedrock.converse span to inspect token counts, latency, and full request/response content

Amazon Bedrock trace span in OpenObserve

Next Steps

Every model call is recorded in OpenObserve. Build dashboards to track token usage over time, set alerts on latency spikes, and correlate Bedrock spans with the rest of your application traces.