Migration & Hybrid
AWS SAA SAA-C03 · Design Resilient Architectures
Migration strategies: the 6 Rs
Migrating to AWS starts with deciding what to do with each application. AWS describes six strategies, known as the 6 Rs: Rehost (“lift and shift” — move the application as-is with minimal changes, the fastest approach); Replatform (“lift, tinker, and shift” — make a few cloud optimizations, such as moving a database to RDS, without changing the core architecture); Repurchase (move to a different product, typically SaaS — replacing a self-hosted CRM with Salesforce, for example); Refactor / Re-architect (reimagine the application using cloud-native features, such as breaking a monolith into microservices — the most effort, but the most benefit); Retire (turn off applications that are no longer needed); and Retain (keep some applications on-premises, for now, because of compliance or technical constraints). A seventh R, Relocate (moving VMware workloads to VMware Cloud on AWS), appears in newer AWS material. The exam tests the 6 Rs by describing a situation and asking which strategy fits: the fastest migration with no code changes is rehost; moving to SaaS is repurchase; rebuilding as microservices is refactor.
Moving data: the right tool for the job
AWS offers a family of data-transfer services, and the exam tests matching the tool to the situation. AWS Database Migration Service (DMS) migrates databases to AWS with minimal downtime, supporting both homogeneous migrations (MySQL to MySQL) and heterogeneous ones (Oracle to PostgreSQL, with AWS SCT for schema conversion). DMS can run a one-time migration or continuous replication, keeping the target in sync until cutover. AWS DataSync moves file data — between on-premises NFS/SMB file systems and AWS storage (S3, EFS, FSx), or between AWS storage services — with scheduling, bandwidth controls, and verification. AWS Storage Gateway is a hybrid storage service that presents AWS storage to on-premises applications as a local file share, iSCSI volume, or virtual tape library, caching frequently used data locally. AWS Transfer Family provides SFTP, FTPS, and FTP endpoints backed by S3 or EFS, for partners and systems that still transfer files over those protocols.
For very large datasets, the network may be the bottleneck, and the answer is physical transfer. The AWS Snow family ships ruggedized devices to your site: Snowcone (small, 8 TB, portable), Snowball Edge (tens of TB, with compute capacity for edge processing), and Snowmobile (an entire truck, up to 100 PB, for exabyte-scale migrations). The exam's rule: if the dataset is huge and the available bandwidth is limited, calculate the transfer time — if it takes months over the network, Snowball is the answer. AWS Application Migration Service (MGN) rehosts servers to AWS by continuously replicating them (the successor to CloudEndure Migration), and AWS Migration Hub provides a central view of migration progress across tools. VM Import/Export converts existing virtual machines into AMIs.
Hybrid architectures
Not every workload moves to the cloud at once, and some never do. Hybrid architectures connect on-premises infrastructure with AWS. The connectivity options (covered in the VPC guide) are Site-to-Site VPN for quick encrypted connections over the internet and Direct Connect for dedicated private bandwidth. AWS Outposts brings AWS infrastructure, services, and APIs into your own data center for workloads that need to run on-premises for latency or data-residency reasons while using the same AWS control plane. Amazon ECS Anywhere and EKS Anywhere run containers on your own hardware with AWS management. AWS Local Zones place AWS compute and storage closer to large population centers for latency-sensitive applications. For DNS, Route 53 Resolver endpoints bridge on-premises and VPC DNS. The exam's hybrid questions usually hinge on one constraint — latency, data residency, or a workload that cannot move — and the answer is the service that satisfies that constraint.
Disaster recovery: RTO, RPO, and the four strategies
Disaster recovery (DR) planning starts with two numbers. Recovery Time Objective (RTO) is how long the business can be down — the maximum acceptable time to restore service. Recovery Point Objective (RPO) is how much data the business can afford to lose — the maximum acceptable age of the data you restore from. A 1-hour RPO means you need backups or replication at least every hour. AWS describes four DR strategies, ordered by increasing cost and decreasing recovery time: Backup and Restore (data is backed up to AWS; on disaster you restore it and rebuild the environment — cheapest, slowest, RTO in hours); Pilot Light (a minimal version of the environment runs in AWS, with data replicated; on disaster you scale it up — RTO in tens of minutes); Warm Standby (a scaled-down but fully functional environment runs in AWS; on disaster you scale it to full size — RTO in minutes); and Multi-Site Active-Active (the full workload runs in two Regions simultaneously with traffic split between them — near-zero RTO, highest cost). The exam tests this by giving an RTO/RPO and a budget and asking for the matching strategy: the tighter the RTO, the warmer the standby must be. AWS Elastic Disaster Recovery (DRS) automates server replication and recovery for the pilot light and warm standby patterns.
Running the migration: waves, cutover, and validation
A migration is a project, not a single event, and AWS's recommended approach is to move in waves: small groups of related applications migrated together, starting with the least complex and least critical, so the team learns the process before tackling the hard systems. Each wave follows the same pattern: assess the applications and their dependencies (AWS Migration Hub and Application Discovery Service help map what talks to what), mobilize the landing zone and tooling, migrate and modernize the wave, and then operate the migrated applications in AWS. AWS Application Migration Service (MGN) is the workhorse for rehosting servers: it continuously replicates on-premises servers (physical or virtual) into AWS, so the cutover window is short — you test the replicated servers, schedule the cutover, and switch over in minutes rather than hours.
Cutover planning is where migrations succeed or fail. A good plan includes a pilot migration of a non-critical application to validate the process end to end, a rollback plan in case the cutover goes wrong, and data validation to confirm that the migrated system holds the same data as the source. For databases, DMS ongoing replication keeps source and target in sync until the moment of cutover, and the cutover itself is usually a short DNS or connection-string change. After cutover, the old system stays available for a defined period in case a rollback is needed, and monitoring is watched closely for performance or cost surprises. The exam tests this discipline through scenario questions: the correct answer is usually the one that minimizes downtime and risk — replicate continuously, test before cutover, and keep a way back — rather than the one that moves everything in a single big-bang event.
Exam traps to watch for
- Confusing RTO with RPO: RTO is how long you can be down (time to restore); RPO is how much data you can lose (age of the last good copy).
- Choosing backup and restore when the RTO is minutes. Backup and restore takes hours; a tight RTO needs warm standby or active-active.
- Using DataSync for a database migration or DMS for file data. DMS moves databases, DataSync moves files.
- Trying to transfer petabytes over a slow internet link. When the math says months, Snowball is the answer.
Practice questions
Test yourself on this guide. Answers are checked instantly, with a full explanation for each question.
Q1.A company is migrating 200 on-premises servers to AWS as quickly as possible with no changes to the applications. Which migration strategy is the company using?
Explanation: Rehost (“lift and shift”) moves applications to AWS without changes, which is the fastest migration strategy and the right description of this scenario. Option C means re-architecting the application for the cloud, which involves significant changes. Option A means replacing the application with a different product (usually SaaS). Option D means turning the application off entirely.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q2.A company needs to migrate a 500 TB dataset to AWS. Its internet connection can sustain 100 Mbps, and the migration must complete within one month. Which solution is MOST appropriate?
Explanation: At 100 Mbps, 500 TB would take over a year to transfer (500 TB is 4,000,000 gigabits; at 0.1 Gbps that is 40,000,000 seconds, or about 463 days) — far beyond the one-month deadline. Snowball Edge devices move the data physically, which is the only option that meets the deadline. Option A fails on the same math. Option B would take weeks just to provision a Direct Connect connection, and even at 1 Gbps the transfer would take about 46 days. Option C accelerates uploads but cannot overcome the 100 Mbps link.
Exam objective: SAA-C03 Task Statement 3.5: Determine high-performing data ingestion and transformation solutions
Q3.A company is migrating its on-premises file server (20 TB of files on NFS shares) to Amazon EFS. The migration must validate that all files were copied correctly, and the company wants to schedule the transfer during off-peak hours. Which AWS service should the solutions architect use?
Explanation: DataSync is purpose-built for moving file data between on-premises NFS/SMB shares and AWS storage services: it schedules transfers, controls bandwidth, and verifies the copied data. Option A migrates databases, not files. Option D is a physical transfer device for very large datasets or limited bandwidth; 20 TB over a normal connection does not need it. Option C presents AWS storage as a local file share for ongoing hybrid access rather than performing a one-time migration with verification.
Exam objective: SAA-C03 Task Statement 3.5: Determine high-performing data ingestion and transformation solutions
Q4.A company runs a legacy application on-premises that it plans to replace with a cloud-based SaaS product during its migration to AWS. Which migration strategy describes this approach?
Explanation: Repurchase means moving to a different product, most commonly replacing a self-hosted application with a SaaS equivalent. Option A moves the existing application unchanged. Option B moves it with minor cloud optimizations. Option D means keeping the application on-premises.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q5.A company's disaster recovery plan requires an RPO of 1 hour. The database is backed up with daily snapshots. What is the problem with this plan?
Explanation: RPO is about data loss: with daily snapshots, the most recent backup could be up to 24 hours old, so a disaster could lose a full day of data. Meeting a 1-hour RPO requires backups or replication at least every hour. Option C confuses the terms: snapshots affect the RPO (data loss), while the time to restore affects the RTO (downtime). Option A is wrong: snapshots are a standard DR mechanism. Option D ignores the gap between 24 hours and 1 hour.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q6.A manufacturing company has an application that must run in its own factory for latency reasons, but the company wants to manage it with the same AWS APIs and tools it uses in the cloud. Which AWS service meets this requirement?
Explanation: AWS Outposts brings AWS infrastructure and services into your own data center, managed through the same AWS console and APIs — the answer for workloads that must run on-premises for latency or data-residency reasons. Option B is a data-transfer and edge-compute device, not a way to run AWS infrastructure permanently in a factory. Option C is a CDN for content delivery. Option A places AWS capacity in metropolitan areas, not inside a customer's own facility.
Exam objective: SAA-C03 Task Statement 3.2: Design high-performing and elastic compute solutions
Q7.A company is migrating an on-premises MySQL database to Amazon RDS for MySQL. The migration must keep the source and target databases in sync until cutover, with minimal downtime. Which AWS service should the solutions architect use?
Explanation: DMS with ongoing (continuous) replication migrates the existing data and then keeps the target in sync with changes on the source until the moment of cutover, minimizing downtime. Option B moves file data, not database contents. Option C physically moves bulk data but does not replicate a live database. Option D replicates S3 objects between buckets and has nothing to do with database migration.
Exam objective: SAA-C03 Task Statement 3.3: Determine high-performing database solutions