Part 2: Amazon Bedrock cost attribution with Amazon Athena and CUDOS
Anthropic
Amazon/AWS
This post demonstrates how to visualize and analyze Amazon Bedrock cost attribution using Amazon Athena queries and CUDOS dashboards. It covers setting up CUR 2.0 with IAM principal data, querying patterns for cost tracking, and new CUDOS capabilities for granular Bedrock cost insights.
The post explains how to set up Cost and Usage Report (CUR) 2.0 through Data Exports with IAM principal data enabled, which populates the line_item_iam_principal column and IAM principal tags. It then shows how to connect CUR to Amazon Athena and provides three SQL query patterns: by IAM principal and usage type, by known IAM principal tags, and dynamic tag discovery using UNNEST. A real-world use case compares costs between a chatbot and a document summarization service. It also introduces CUDOS dashboard version 5.8, which includes a Bedrock section with flexible grouping, cost-per-million-tokens tracking, and interactive drill-down filtering. Athena queries cost $5 per TB scanned with a 10 MB minimum, so scoping queries to a single month keeps costs low.
- Abbreviations
- AWS = Amazon Web Services — Amazon Web Services
- CUR = Cost and Usage Report — Отчет о затратах и использовании
- IAM = Identity and Access Management — Управление доступом и идентификацией
- S3 = Simple Storage Service — Простой сервис хранения
- SQL = Structured Query Language — Структурированный язык запросов
- CUDOS = Cost and Usage Dashboard Operational System — Система операционного управления затратами и использованием
- CID = Cloud Intelligence Dashboards — Облачные интеллектуальные панели мониторинга
- IaC = Infrastructure as Code — Инфраструктура как код
- CLI = Command Line Interface — Интерфейс командной строки
- TB = Terabyte — Терабайт
Source: AWS ML blog —
original