Building a Cross-Account Central Model Catalog with SageMaker Model Registry and Model Group Resource Policies
A company needs to create a central catalog for all the company's ML models. The models are in AWS accounts where the company developed the models initially. The models are hosted in Amazon Elastic Container Registry (Amazon ECR) repositories. Which solution will meet these requirements?
Community Votes
100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.
Community Insight
SageMaker Model Registry provides a true model catalog with model groups for organizing models, and a cross-account resource policy attached to each model group lets another account reference those models, which is how a central account aggregates a catalog without moving the model artifacts.
A company needs a central catalog for all of its ML models, which live in the AWS accounts where they were originally developed and are hosted in ECR repositories. The catalog has to span accounts, which means the solution has to work across account boundaries rather than within one.
Configuring ECR cross-account replication to gather the models, which copies container images rather than building a model catalog. A model catalog needs metadata, organization, and discoverability, and an ECR repository listing provides neither.
Community Discussion (3 comments)
Comments & Corrections
No comments yet — spotted an error or have a note? Share it below.
Expert Analysis
Why the Answer Is Correct
The requirement is a central catalog for ML models that are spread across the AWS accounts where they were developed and hosted in ECR. SageMaker Model Registry is the service built for cataloging ML models, and creating a model group for models hosted in ECR registers them with the metadata and organization a catalog requires. Putting the Model Registry in a new central account and attaching a cross-account resource policy to each model group in the original accounts is what allows that central registry to reference the models across account boundaries without relocating the artifacts. The vote was 88 for C. GiorgioGss identified the decisive point that the question asks for a central catalog, which implies metadata and lineage, and noted that ECR cannot serve as that solution.Why the Other Options Are Wrong
Configuring ECR cross-account replication so each model is visible in every account (A) replicates container images between repositories, and a list of replicated images is not a model catalog, so it provides no model metadata, grouping, or discoverability. Creating a new account with a new ECR repository as the central catalog and replicating into it (B) is the same category error as option A in a different shape, and a4002bd argued that it centralizes models in a single ECR repository, which is repository consolidation rather than cataloging. Using an AWS Glue Data Catalog with a crawler to migrate the models and cross-account access (D) is a mismatch because the Glue Data Catalog catalogs data files and schemas, not ML models, and a crawler enumerating ECR repositories does not produce model metadata.Community Comment Notes
The community favored C at 88 votes, and GiorgioGss gave the reasoning that decided the question, observing that a central catalog implies metadata and lineage tracking are included, and that ECR could not be the solution. aws_Tpozit expanded on the same logic in detail, matching the keywords of a central catalog across accounts with the Model Registry model group plus cross-account resource policy. The single dissenting comment, from a4002bd, argued for option B on the grounds that cross-account replication centralizes the models in one repository, which is a reasonable reading of repository consolidation but does not satisfy the catalog requirement the question states.Official Reference
Related Analysis
Practice All MLA-C01 Questions
Access 115 questions with complete answers and detailed explanations.
View Full MLA-C01 Practice Test →