improve architecture doc

aws-solutions-library-samples · Jan 30, 2025 · 89c4ef4 · 89c4ef4
1 parent bccb695
commit 89c4ef4
Show file tree

Hide file tree

Showing 7 changed files with 429 additions and 6 deletions.
diff --git a/docs/_static/drawio/sdlf-architecture-datalake.drawio b/docs/_static/drawio/sdlf-architecture-datalake.drawio
diff --git a/docs/_static/drawio/sdlf-architecture-datamesh.drawio b/docs/_static/drawio/sdlf-architecture-datamesh.drawio
diff --git a/docs/_static/sdlf-architecture-datalake.png b/docs/_static/sdlf-architecture-datalake.png
diff --git a/docs/_static/sdlf-architecture-datamesh.png b/docs/_static/sdlf-architecture-datamesh.png
diff --git a/docs/_static/sdlf-architecture.png → docs/_static/sdlf-pipeline-full.png b/docs/_static/sdlf-architecture.png → docs/_static/sdlf-pipeline-full.png
diff --git a/docs/architecture.md b/docs/architecture.md
@@ -4,21 +4,29 @@ SDLF supports both a centralized datalake deployment pattern and decentralized d
 
 ## Centralized Data Lake
 
-![Centralized Data Lake Architecture](_static/sdlf-architecture.png)
+![Centralized Data Lake Architecture](_static/sdlf-architecture-datalake.png)
 
 !!! warning
-    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF Implementation
+    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF implementation.
 
 ## Data Mesh
 
-![Data Mesh Architecture](_static/sdlf-architecture.png)
+The Data Mesh pattern is fundamentally about decentralized data ownership, with data owned by specialized domain teams rather than a centralized data team. This usually means:
+- each data domain team has its own dedicated data infrastructure, for production and/or consumption
+- each data domain team is able to deploy the tools and infrastructure it needs - a self-serve data platform
+
+A governance layer is federating data assets in a business catalog to ensure compliance against policies and standards, and ease of data sharing across teams.
+
+As such, it can be seen as a collection of data domain-specific datalakes deployed with SDLF. [Amazon SageMaker Data and AI Governance](https://aws.amazon.com/sagemaker/data-ai-governance/) (built on Amazon DataZone) can be used for the governance layer.
+
+![Data Mesh Architecture](_static/sdlf-architecture-datamesh.png)
 
 !!! warning
-    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF Implementation
+    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF implementation.
 
 ## Transactional Data Lake
 
 Using [Iceberg](https://docs.aws.amazon.com/prescriptive-guidance/latest/apache-iceberg-on-aws/introduction.html).
 
 !!! warning
-    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF Implementation
+    We strongly recommend that customers conduct a [Well Architected Review](https://aws.amazon.com/architecture/well-architected/) of their SDLF implementation.
diff --git a/docs/constructs/pipeline.md b/docs/constructs/pipeline.md
@@ -13,7 +13,7 @@ Each pipeline is divided into stages (i.e. StageA, StageB...), which map to AWS
 
 Each Step Functions is comprised of one or more steps relating to operations in the orchestration process (e.g. Starting an Analytical Job, Running a crawler...).
 
-![SDLF Architecture](../_static/sdlf-architecture.png)
+![SDLF Architecture](../_static/sdlf-pipeline-full.png)
 
 An example architecture for a SDLF pipeline is detailed in the diagram above. The entire process is event-driven.