How Do Outposts and Wavelength Meet Data Residency and Low Latency?
A multinational healthcare company is building a GenAI solution to analyze patient medical records and clinical trial data to identify treatment patterns and generate research insights. Some of the countries where the company operates require the company to store patient data on premises within each country. The solution must provide low-latency, AI-assisted diagnostics for remote clinics around the world. The solution must use cloud-based foundation models (FMs) to perform advanced, large-scale analytics. The company needs a solution that ensures data residency, delivers sub-100 ms inference at clinics, and supports consistent development workflows across hybrid environments. Which solution will meet these requirements with the LEAST operational overhead?
Community Insight
The exam tests whether you can split a hybrid GenAI workload between on-premises compute (Outposts, for residency) and telco edge zones (Wavelength, for latency) rather than forcing every requirement through a single centralized or self-managed EC2 fleet.
This question tests hybrid GenAI architecture design for healthcare data residency plus sub-100 ms edge inference, and it establishes AWS Outposts combined with AWS Wavelength as the pattern that satisfies both with the least operational overhead.
Learners most often pick B because Amazon EC2 in each country appears to satisfy residency, but self-hosting open source FMs on EC2 in every country and manually exporting CSV files creates huge operational overhead and no real low-latency inference path at the clinics.
Community Discussion (1 comments)
Comments & Corrections
No comments yet — spotted an error or have a note? Share it below.
Expert Analysis
Why the Answer Is Correct
Option D maps each hard requirement to the AWS service purpose-built for it: AWS Outposts keeps patient records physically on-premises inside each country, which directly satisfies the on-premises data residency mandate without replicating PHI off-site. AWS Wavelength embeds compute and storage inside telco networks at the edge, so remote clinics reach a nearby inference endpoint and can realistically achieve the sub-100 ms latency target. Both are fully managed AWS infrastructure, so model deployment pipelines, IAM, and control-plane tooling stay consistent with the company's existing cloud workflows, minimizing operational overhead while still allowing cloud-based FMs for advanced analytics. This is exactly the hybrid pattern the question is probing.Why the Other Options Are Wrong
Option A violates residency outright by replicating on-premises patient data to Amazon S3 and by funneling all requests through a single centralized inference endpoint, which also makes the sub-100 ms target impossible for clinics far from that Region. Option C is worse still: it replicates patient data to the cloud (breaking residency) while centralizing training and inference (breaking latency), and it provides no on-premises enforcement. Option B technically respects residency but shifts heavy burden onto the customer, since EC2 GPU fleets, Systems Manager patching of open source models in every country, and manual CSV exports for cross-country analysis are all high-overhead, error-prone processes that also fail the low-latency and consistent-workflow goals.Community Comment Notes
The single community comment, from TNT87, selects D, reasoning that "Use AWS Outposts to process and store sensitive on-premises patient data" combined with AWS Wavelength covers both requirements. That matches the independent analysis here: Outposts addresses residency and Wavelength addresses edge latency, so the community input is aligned rather than a vote-only majority.Official Reference
Exam Strategy
When a question names two separate hard constraints such as data residency and sub-100 ms latency, look for an option that assigns a different purpose-built service to each constraint instead of one service trying to do everything. Eliminate any option that replicates regulated data off-premises or that relies on manual file exports, since both signal high operational overhead.
Frequently Asked Questions
Why is deploying open source FMs on EC2 in each country not the least-overhead choice?
EC2 forces you to build and patch GPU fleets, orchestrate model updates per country, and manually move insights for analysis, creating far more operational work than managed Outposts and Wavelength.
Why does a centralized inference endpoint fail the sub-100 ms requirement?
A single Region-hosted endpoint adds cross-continent round-trip time for remote clinics, so latency targets cannot be met regardless of any Global Accelerator improvement.
Related Analysis
Practice All AIP-C01 Questions
Access 110 questions with complete answers and detailed explanations.
View Full AIP-C01 Practice Test →