AWS Security ChangesHomeSearch

AWS wellarchitected: Rewrite ADVSUS05 as trust, safety, and privacy workload efficiency

Service: wellarchitected · 2026-11-11 · Documentation low

File: wellarchitected/latest/video-streaming-advertising-lens/advsus05.md

Summary

Replaces the 'Data management' sustainability question with 'Trust, safety, and privacy workload efficiency', reframing ADVSUS05 around right-sizing compute for privacy-collaboration, fraud-detection, and content-moderation workloads, batching non-real-time work, analyzing data where it resides to avoid cross-Region movement, and tiering/retaining signal, audit, and result data only as long as compliance, security, or model accuracy requires. Best-practice links and the related ADVSUS04-BP02 title are also updated.

Security assessment

The change is sustainability guidance (energy efficiency, right-sizing, batching, data locality) that only tangentially touches security concepts such as data retention, audit trails, and cross-Region data movement. It documents no security feature, control, or vulnerability fix, and no CVE or incident is referenced, so it is a low-severity, security-adjacent documentation rewrite.

Evidence

+  * Retaining raw signals, results, and audit trails indefinitely on a single high-performance storage tier with no lifecycle or deletion policy. 

Diff

diff --git a/wellarchitected/latest/video-streaming-advertising-lens/advsus05.md b/wellarchitected/latest/video-streaming-advertising-lens/advsus05.md
index bd7f329..77af8b2 100644
--- a//wellarchitected/latest/video-streaming-advertising-lens/advsus05.md
+++ b//wellarchitected/latest/video-streaming-advertising-lens/advsus05.md
@@ -7 +7 @@
-# Data management
+Capability intentCommon issues to watch for
@@ -9 +9,5 @@
-ADVSUS05: How are you optimizing data storage and retrieval?  
+# Trust, safety, and privacy workload efficiency
+
+Privacy collaboration, fraud detection, and content moderation are the data-intensive trust-and-safety workloads that run alongside advertising delivery. Each analyzes large volumes of impression, click, content, or partner data on an ongoing basis, so each carries a resource and carbon footprint that grows with advertising activity. The sustainability levers are shared across all three: scale compute to actual demand rather than provisioning for peak, run ongoing work on energy-efficient processors, defer and batch work that has no real-time requirement, and analyze data where it resides to avoid moving it across Regions. 
+
+ADVSUS05: How do you minimize the resource and carbon impact of privacy-collaboration, fraud-detection, and content-moderation workloads?   
@@ -10,0 +15,24 @@ ADVSUS05: How are you optimizing data storage and retrieval?
+[ADVSUS05-BP01 Optimize privacy workload processing patterns and resource allocation for sustainability](./advsus05-bp01.html)  
+[ADVSUS05-BP02 Optimize fraud detection systems for resource efficiency](./advsus05-bp02.html)  
+[ADVSUS05-BP03 Optimize content moderation systems for sustainable operation](./advsus05-bp03.html)  
+  
+## Capability intent
+
+  * Privacy collaboration, fraud detection, and content moderation run on right-sized, energy-efficient compute that scales with actual demand rather than remaining provisioned for peak. 
+
+  * Machine learning inference and other continual processing run on purpose-built, high-efficiency hardware selected for performance per watt. 
+
+  * Work without a real-time latency requirement is separated from latency-sensitive paths, then batched and scheduled to raise utilization and match periods of lower grid carbon intensity. 
+
+  * Data is analyzed where it resides and aggregated close to its source, limiting cross-Region movement of large impression, click, and media datasets. 
+
+  * Signal, audit, and result data is stored on tiers matched to its access pattern and retained only as long as compliance, security, or model accuracy requires. 
+
+
+
+
+## Common issues to watch for
+
+  * Running privacy, fraud, or moderation workloads continuously on persistent, over-provisioned instances regardless of actual demand. 
+
+  * Processing every event or asset at full fidelity and in real time when intelligent sampling or batching would meet the same business and risk requirement. 
@@ -12 +40 @@ ADVSUS05: How are you optimizing data storage and retrieval?
-Data management involves provisioning the minimum amount of storage required to meet your workload's needs. For existing storage, understand the needs and usage of your data to set appropriate lifecycles for your data to move data to the most efficient type of storage. Delete unnecessary data, when possible, to reduce storage footprint. 
+  * Defaulting to general-purpose x86 compute without evaluating energy-efficient, purpose-built processor options for matching, detection, and inference. 
@@ -14 +42 @@ Data management involves provisioning the minimum amount of storage required to
-###### Best practices
+  * Centralizing analysis in a single Region and moving large volumes of impression, click, or media data across Regions, or copying partner datasets rather than analyzing them where they reside. 
@@ -16 +44 @@ Data management involves provisioning the minimum amount of storage required to
-  * [ADVSUS05-BP01 Identify and remove redundant data across storage](./advsus05-bp01.html)
+  * Retaining raw signals, results, and audit trails indefinitely on a single high-performance storage tier with no lifecycle or deletion policy. 
@@ -27 +55 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please
-ADVSUS04-BP02 Use serverless transaction processing 
+ADVSUS04-BP02 Match model capability to task complexity to reduce agent fleet energy use
@@ -29 +57 @@ ADVSUS04-BP02 Use serverless transaction processing
-ADVSUS05-BP01 Identify and remove redundant data across storage
+ADVSUS05-BP01 Optimize privacy workload processing patterns and resource allocation for sustainability