Most SAP-C02 questions live entirely inside a Region — VPCs, subnets, Availability Zones, managed services. Then a scenario arrives that breaks the frame: a factory floor that cannot tolerate a round trip to the nearest Region, a country that legally requires customer data to stay on national soil, a 5G application that needs single-digit-millisecond latency to mobile users, or fifty petabytes that would take months to push over the wire. These are hybrid and edge problems, and AWS built a specific family of services to answer them — Outposts, Local Zones, Wavelength, and the Snow Family — each solving a different constraint.
The SAP-C02 exam does not ask you to configure these services. It asks you to choose between them under pressure, given a scenario with a latency target, a data-residency rule, a connectivity limitation, or a physical-location requirement. Pick Local Zones when the answer is Outposts and you lose the mark. This guide gives you the decision framework the professional exam expects: what each service is, the one problem it exists to solve, and the trade-offs that separate the right answer from the plausible-looking distractor.
If you want the full exam map first, start with the AWS Solutions Architect Professional guide for 2026 and sequence your prep with the SAP-C02 study plan. This piece assumes you already understand AWS networking at the VPC and Transit Gateway level — hybrid and edge build directly on top of that foundation.
The Four Problems, The Four Services
Before the detail, anchor the whole topic on a single mapping. Every hybrid/edge scenario on the exam reduces to one of these constraints, and each maps to a primary service:
| The constraint in the scenario | Primary service | One-line reason |
|---|---|---|
| Workload must run on-premises (data residency, ultra-low latency to local systems, gradual migration) | AWS Outposts | Native AWS hardware in your data center, same APIs. |
| Users in a metro area far from any Region need low latency, still on AWS-owned sites | Local Zones | AWS-operated infrastructure closer to population centers. |
| Mobile / 5G end users need the lowest possible latency over the carrier network | AWS Wavelength | Compute embedded inside telecom providers’ 5G networks. |
| Move large data into AWS, or run compute in a disconnected/rugged location | Snow Family | Physical devices for edge compute and bulk data transfer. |
Memorize this table and 80% of the exam’s hybrid/edge questions become a matter of reading the scenario for the constraint and reading it back off the row. The rest of this guide fills in the trade-offs that separate closely-related options — because the exam’s favorite trick is offering two services from this table and making you justify one over the other.
AWS Outposts: The Region in Your Data Center
Outposts is a rack (or a single 1U/2U server, in the smaller form factor) of genuine AWS-designed hardware that AWS installs and operates inside your own facility. It runs a subset of AWS services — EC2, EBS, ECS/EKS, RDS, S3 on Outposts, and more — with the same APIs, same console, and same tooling as the parent Region. To your teams, an Outpost looks like an extension of a VPC that happens to be sitting in your building.
You reach for Outposts when the workload genuinely cannot live in a Region:
- Low-latency to on-premises systems. A manufacturing line or trading system that must talk to local equipment or databases with microsecond-to-low-millisecond latency cannot afford a round trip to the Region.
- Data residency and local processing. Regulations or contracts require certain data to be processed and stored physically on-premises, while you still want the AWS operating model.
- Gradual migration. You are modernizing incrementally and want AWS services running next to legacy systems during the transition.
The connectivity model matters for the exam. An Outpost is anchored to a parent Region over a Service Link — a reliable connection (typically over Direct Connect or a VPN) back to the Region’s control plane. Local workloads on the Outpost keep running even if the Service Link drops (existing instances stay up), but you lose the ability to launch/manage resources and reach Region-only services until it recovers. When a scenario stresses “must keep operating during a network outage to AWS,” remember Outposts survives for running workloads but depends on the Region for management.
Exam tells for Outposts: the words on-premises, in the customer’s data center, data must not leave the building, consistent hybrid experience with the same AWS APIs, or ultra-low latency to existing local equipment.
Local Zones: AWS Infrastructure Closer to Your Users
A Local Zone is an extension of a Region placed in a large metro area that is geographically far from the Region’s main Availability Zones. It is AWS-owned and AWS-operated — you are not putting hardware in your building — but it sits physically closer to a population center, so single-digit-millisecond latency to users in that metro becomes possible.
You associate a Local Zone by extending your VPC into it (a subnet in the Local Zone) and launching a supported subset of services — EC2, EBS, load balancing, and others — right there. Traffic to end users in that metro stays local; anything else routes back to the parent Region over AWS’s backbone.
The distinction the exam loves is Outposts vs Local Zones:
| Dimension | Outposts | Local Zones |
|---|---|---|
| Who owns/hosts the hardware | You host it in your facility | AWS owns and operates the site |
| Primary driver | On-prem residency, latency to your systems | Latency to end users in a metro AWS doesn’t have a Region in |
| You manage physical space/power? | Yes (space, power, cooling in your DC) | No |
| Typical scenario | Factory, hospital, on-prem database next door | Media/gaming/real-time app for a distant city |
Read the scenario for whose building the compute sits in. If the requirement is “in our data center” or “on-premises,” it is Outposts. If it is “our users in [distant city] see too much latency and we don’t need it on-prem,” it is a Local Zone.
Exam tells for Local Zones: end users in a specific metro, no Region nearby, low-latency without managing on-prem hardware, real-time gaming/media/AR-VR for a city far from the Region.
AWS Wavelength: Compute Inside the 5G Network
Wavelength goes one step further out than Local Zones. It places compute and storage inside telecommunications providers’ 5G networks, at the edge of the mobile carrier’s infrastructure (a Wavelength Zone). Traffic from a mobile device reaches your application without leaving the carrier’s network to traverse the public internet, which is what makes ultra-low latency to mobile users achievable.
The defining constraint is the audience: mobile end users on 5G. Think connected vehicles, live mobile video/AR, real-time mobile gaming, or industrial IoT over cellular. If the scenario is not fundamentally about mobile/5G users, Wavelength is almost certainly a distractor.
The mental ladder of “how close to the user” is worth internalizing:
- Region → users near the Region, standard case.
- Local Zone → users in a metro far from the Region, over the internet but on nearby AWS infrastructure.
- Wavelength → mobile users, compute embedded in the 5G network so traffic never hits the public internet.
- Outposts → the compute is on-premises with the customer’s own systems.
Exam tells for Wavelength: 5G, mobile devices, carrier network, ultra-low latency for a mobile app, connected cars / mobile AR.
The Snow Family: Data Transfer and Disconnected Edge Compute
The Snow Family solves two related problems: moving large volumes of data into (or out of) AWS when the network is too slow, and running compute at the edge in locations that are rugged, bandwidth-constrained, or disconnected entirely.
- AWS Snowcone — the smallest, most portable device; a few terabytes of usable storage, suitable for edge data collection and light compute in the field (drones, vehicles, remote sites). Can ship data back or transfer online via AWS DataSync.
- AWS Snowball Edge — larger devices in two flavors: Storage Optimized (bulk data migration, tens of terabytes) and Compute Optimized (EC2 instances and even GPU options for edge processing where connectivity is poor).
- AWS Snowmobile — historically an exabyte-scale option (a shipping container on a truck) for massive one-time migrations. (Note: for the exam, know it exists as the answer to “we must move many petabytes to exabytes and the network would take years”; AWS has deemphasized it, so a Snowball fleet is the more common modern answer for large-but-not-exabyte migrations.)
The exam’s key decision is Snow vs network transfer. Do the arithmetic mentally: pushing tens or hundreds of terabytes over a saturated or low-bandwidth link can take weeks to months, during which the data keeps growing. When the scenario mentions limited bandwidth, a hard deadline, or a disconnected/harsh location, physical shipment via Snow beats the wire.
Contrast Snow with the online transfer services so you don’t confuse them:
| Need | Service |
|---|---|
| Large one-time bulk transfer, poor/no connectivity, deadline | Snowball / Snowcone (physical) |
| Ongoing scheduled transfer over the network to S3/EFS/FSx | AWS DataSync |
| Continuous streaming/hybrid file access to S3 | Storage Gateway |
| Edge compute in a disconnected location | Snowball Edge Compute Optimized |
Exam tells for Snow: petabytes, limited/no internet, ship the data, ruggedized, disconnected edge, would take months over the network.
Hybrid DNS: Making Names Resolve Both Ways
Once you have workloads split across AWS and on-premises, one problem shows up in nearly every hybrid scenario: name resolution across the boundary. On-prem servers need to resolve AWS private hosted zone records; AWS resources need to resolve on-prem internal names. The service that bridges this is Route 53 Resolver, using two kinds of endpoints:
- Inbound endpoints let on-premises systems query Route 53 (and your private hosted zones) — DNS queries flow into the VPC from your data center.
- Outbound endpoints with resolver rules forward queries from AWS out to your on-premises DNS servers for specific domains — so an EC2 instance can resolve
db.corp.internalagainst your internal DNS.
The direction is the whole point, and the exam tests exactly this:
| Requirement | Endpoint |
|---|---|
| On-prem hosts must resolve names in an AWS private hosted zone | Inbound resolver endpoint |
| AWS resources must resolve on-premises internal domain names | Outbound resolver endpoint + forwarding rule |
| Both directions | Both endpoints |
Pair this with the connectivity layer — Direct Connect or Site-to-Site VPN carrying the DNS traffic — and you have the complete hybrid DNS picture. When a scenario says “instances in the VPC can’t resolve our corporate .internal domain,” the answer is an outbound Resolver endpoint with a forwarding rule pointing at the on-prem DNS servers.
Putting It Together: A Decision Walkthrough
Here is how a professional-level question actually plays out. Scenario: “A media company streams live, low-latency video to viewers in a large city 1,500 km from the nearest AWS Region. They do not want to operate any hardware themselves. Which approach minimizes latency?”
Walk the framework:
- What’s the constraint? Latency to end users in a distant metro. Not on-prem. Not mobile-5G-specific. Not data movement.
- Whose hardware? They explicitly don’t want to run any — rules out Outposts.
- Mobile 5G? Not stated as the defining factor — Wavelength is a distractor here.
- Answer: Local Zone in that metro, extending the VPC to serve viewers locally.
Change one word — “viewers on their 5G mobile app” — and the answer shifts to Wavelength. Change it to “must run inside our own broadcast facility for latency to studio equipment” and it becomes Outposts. This sensitivity to a single phrase is the entire skill the exam is testing, and it is why rote memorization of features loses to a clean decision framework.
Practice the Judgment, Not Just the Facts
You can read every service page on the AWS site and still miss these questions, because the exam is not testing recall — it is testing whether you can extract the deciding constraint from a wall of scenario text and map it to the one right service while three plausible distractors sit next to it. That judgment only comes from working realistic scenario questions and reading the explanations for why the distractors are wrong.
That is what full-length practice is for. The AWS Solutions Architect Professional (SAP-C02) Mock Exam Bundle gives you eight mock exams of scenario-style questions with detailed explanations, so you build the reflex of spotting “on-premises” vs “distant metro” vs “5G mobile” vs “ship the data” and choosing correctly under the clock. Combine it with the SAP-C02 networking deep dive for the connectivity layer and how to pass SAP-C02 on your first attempt for the exam-day strategy.
Frequently Asked Questions
What is the difference between AWS Outposts and Local Zones?
Outposts is AWS hardware installed in your own data center — you host it, and you use it when the workload must run on-premises for residency or ultra-low latency to local systems. A Local Zone is AWS-owned infrastructure in a metro area far from a Region, used to reduce latency for end users in that city without you managing any hardware. The deciding question: does the compute need to sit in your building (Outposts) or just closer to your users (Local Zone)?
When should I use AWS Wavelength instead of a Local Zone?
Use Wavelength when the low-latency requirement is specifically for mobile users on 5G, because Wavelength places compute inside the carrier’s network so traffic never traverses the public internet. Local Zones serve users in a metro over standard internet paths. If the scenario centers on mobile/5G, choose Wavelength; otherwise, a Local Zone is usually the intended answer.
How do I choose between the Snow Family and network transfer like DataSync?
Estimate transfer time. If moving the data over the available bandwidth would take weeks or months, or the location has poor/no connectivity or a hard deadline, use a physical Snow device. If you have adequate bandwidth and need ongoing, scheduled transfers to S3/EFS/FSx, use DataSync. Snow is for large one-time bulk moves and disconnected edge compute; DataSync is for repeatable online transfers.
How does DNS resolution work in a hybrid AWS architecture?
Use Route 53 Resolver endpoints. An inbound endpoint lets on-premises systems resolve names in AWS private hosted zones. An outbound endpoint, paired with forwarding rules, lets AWS resources resolve on-premises internal domains by forwarding those queries to your on-prem DNS servers. Deploy both when resolution must work in both directions, carried over Direct Connect or a Site-to-Site VPN.
Do Outposts keep working if the connection to AWS goes down?
Existing workloads running on an Outpost continue to operate through a temporary loss of the Service Link to the parent Region. However, you cannot launch or manage resources, and Region-dependent services are unavailable, until connectivity is restored. So Outposts survives an outage for running workloads but depends on the Region for control-plane operations — an important nuance when a scenario stresses offline resilience.
Is AWS Snowmobile still the answer for very large migrations?
Snowmobile was the exabyte-scale, truck-based option for massive one-time migrations. For the exam, recognize it as the answer to “move exabytes when the network would take years,” but note AWS has deemphasized it; for large-but-not-exabyte migrations, a fleet of Snowball Edge devices is the more common modern recommendation.
Conclusion
Hybrid and edge is a small topic with an outsized payoff on the SAP-C02, because the questions are almost always winnable once you have the decision framework: Outposts for compute on-premises, Local Zones for users in a distant metro, Wavelength for mobile 5G users, and the Snow Family for bulk data movement and disconnected edge compute — with Route 53 Resolver endpoints stitching DNS across the boundary. The exam’s difficulty is not in the services themselves but in reading the scenario for the one deciding constraint and resisting the plausible distractor.
Build that judgment through reps. Follow the SAP-C02 study plan, keep the Solutions Architect Professional guide for 2026 as your map, and drill scenario questions with the SAP-C02 mock exam bundle until “on-premises,” “distant metro,” “5G mobile,” and “ship the data” trigger the right service automatically.