How BMW Group Detects Cost Anomalies Across 14,000 Cloud Accounts

Serdar HocamAuthor & Editor

BMW Group actively tracks thousands of cloud accounts daily thanks to a custom FinOps system developed with Reply on AWS infrastructure.

◉ 0 views
How BMW Group detects cost anomalies across 14,000 cloud accounts | Amazon Web Services

BMW Group actively utilizes an internal FinOps system called Cloud Efficiency Analytics, developed in partnership with Reply on AWS, to monitor 14,000 cloud accounts across the company.

Cloud Efficiency Analytics System

BMW Group operates an internal FinOps system named Cloud Efficiency Analytics to audit 14,000 company-wide cloud accounts.

Processing Billing Data

The system collects billing data comprising approximately 3 billion rows and 500 columns on a monthly basis from AWS Cost and Usage Reports and other providers.

Cost Forecasting with Meta Prophet

To establish cost baselines, Meta's open-source forecasting library Prophet is used, examining the daily history for each account-service pair.

Advantages of Serverless Architecture

Thanks to AWS Step Functions and Lambda functions, the daily scan of 14,000 accounts is completed in approximately 20 minutes at low costs.

Alert Mechanisms and Filtering

To reduce and manage false alarms, various filtering layers are applied, such as low-spend exclusions and custom deviation thresholds.

Root Cause Analysis and Dashboard

Account owners receive email notifications for detected anomalies and can perform detailed analysis via a dashboard on Amazon QuickSight.