Observability and Billing Transparency for Vertex AI RAG Engine

Google Cloud Build with AI Agents
1

Hi Everyone,

I am currently using the Vertex AI RAG Engine with the default managed Spanner option (RagManagedDb). While the service works well, I am struggling with monitoring and resource management. I would appreciate some clarity on the following points:

  1. Billing & Cost Tracking: I cannot find any specific line items for RagManagedDb in my Google Cloud Billing reports. Even when using free credits, I expected to see a usage entry (even with a $0.00 amount). Are these costs bundled into general Vertex AI SKUs? How can I isolate the costs associated specifically with RAG Engine storage and operations?

  2. Storage Visibility: Is there a way to check the actual storage consumption (GB) of my corpora and indexes within the managed Spanner instance? Without this information, capacity planning and cost forecasting are nearly impossible.

  3. Performance Metrics: Can we access any Spanner-level metrics (e.g., CPU utilization, API latency, or I/O) during heavy file uploads or retrieval tasks? I need these for diagnostic purposes and to optimize our ingestion pipelines.

I understand that (RagManagedDb) is a fully managed service where Google handles the underlying infrastructure, but as an enterprise user, I need basic observability into the resources I am consuming.

Has anyone found a way to extract this data via Cloud Monitoring (Stackdriver) or specific API calls?

Thanks in advance for your help!