AWS wellarchitected documentation change

Service: wellarchitected · 2026-01-28 · Documentation low

File: wellarchitected/latest/financial-services-industry-lens/fsicost09.md

Summary

Added data lifecycle policies for RAG artifacts and AI datasets across storage tiers

Security assessment

Changes focus on cost-optimized storage tiering for AI artifacts using S3 lifecycle policies. While data storage is mentioned, there's no discussion of encryption, access controls, or data protection - purely cost-driven archival strategies. No security implications identified.

Diff

diff --git a/wellarchitected/latest/financial-services-industry-lens/fsicost09.md b/wellarchitected/latest/financial-services-industry-lens/fsicost09.md
index 43c16d5a3..2f38f86a4 100644
--- a//wellarchitected/latest/financial-services-industry-lens/fsicost09.md
+++ b//wellarchitected/latest/financial-services-industry-lens/fsicost09.md
@@ -13 +13,3 @@ FSI companies usually have long retention policies for their regulatory and audi
-FSI companies usually have long retention policies for their regulatory and audit requirements. They usually span multiple years and might even be able to take up to a day or two to be able to retrieve old data. Defining data retention policies and corresponding architecture to transfer data from main storage to archival storage is important. This can be achieved by transferring data from RDS database to S3, or creating snapshot and storing it for better cost efficiencies. 
+FSI companies usually have long retention policies for their regulatory and audit requirements. They usually span multiple years and might even be able to take up to a day or two to be able to retrieve old data. Defining data retention policies and corresponding architecture to transfer data from main storage to archival storage is important. This can be achieved by transferring data from RDS database to S3 or creating a snapshot and storing it for better cost efficiencies. 
+
+Apply lifecycle policies for Retrieval-Augmented Generation (RAG) artifacts and generative AI datasets: maintain hot vector indexes (for current-quarter documents or active knowledge bases) in high-performance vector databases, transition warm data (historical embeddings or older training data) to object storage such as Amazon S3 Standard–IA, and archive cold or infrequently accessed corpora (for example, legacy PDFs, or processed embeddings) in Amazon Glacier or Deep Archive. Automate transitions using S3 Lifecycle policies to minimize long-term storage costs while preserving retrieval fidelity when needed.