Migration and Innovation

CAF perspectives, the 7 Rs of migration, data transfer tools, and AWS AI/ML services.

What is it?

AWS Cloud Adoption Framework (CAF) helps organizations plan their cloud journey through six perspectives: Business, People, Governance, Platform, Security, and Operations. The first three focus on business capabilities, the last three on technical ones.

The 7 Rs are strategies for moving each application:

  • Retire: switch it off; nobody needs it.
  • Retain: leave it where it is for now.
  • Rehost (lift and shift): move it as is to EC2.
  • Relocate: move at the hypervisor level (for example VMware workloads) with little change.
  • Repurchase: replace with a SaaS product.
  • Replatform (lift, tinker and shift): small optimisations such as moving to a managed database.
  • Refactor / re-architect: redesign to use cloud-native features.

Moving data and servers

  • AWS Snow Family (Snowball Edge devices): ruggedised appliances to move large amounts of data or run edge compute when networks are slow. Snowmobile (the truck-sized option) has been retired, and some smaller Snow devices have been discontinued for new customers, so check availability.
  • AWS DataSync: automates online transfers between on-premises storage and AWS storage.
  • AWS Application Migration Service (MGN): replicates servers into AWS for rehosting. Database Migration Service handles databases.

Innovation with AI/ML

  • Amazon SageMaker: build, train and deploy your own machine learning models.
  • Amazon Bedrock: access to foundation models through an API to build generative AI applications.
  • Amazon Lex: conversational chatbots.
  • Amazon Transcribe: speech to text.
  • Amazon Polly: text to speech.
  • Amazon Comprehend: text analysis (sentiment, entities).
  • Amazon Textract: extract text and structure from scanned documents.
  • Amazon Rekognition: image and video analysis.
  • Amazon Translate: language translation.
  • Amazon Kendra and similar services: intelligent search.

Benefits of the CAF: following its perspectives brings more reliable cloud migrations, higher business value, lower risk, better alignment between business and IT, and clearer skills and roles across the organisation.

Discovery and tracking tools: AWS Application Discovery Service collects server inventory, utilisation and dependencies from on-premises data centers; AWS Migration Hub gives one place to track migration progress across tools; AWS Application Migration Service (MGN) performs the lift-and-shift replication; AWS Transfer Family moves files over SFTP/FTPS; DataSync, DMS and the Snow Family move data. Use the AWS Migration Evaluator to build a business case.

Explain like I'm 10

Migrating is like relocating a large household. Retire is donating the broken lamp, retain is keeping the piano in the old house for now, rehost is moving the sofa exactly as it is, replatform is swapping the heavy fridge for a built-in one, repurchase is buying a rental subscription instead of a washing machine, and refactor is redesigning the kitchen. Snowball is the container you fill and have shipped when the moving van would take forever.

Examples

Classifying a portfolio with the 7 Rs

Application              Decision      Why
-----------------------  ------------  -----------------------------------------
Old intranet wiki        Retire        Usage near zero
Mainframe billing        Retain        High risk; revisit in two years
Windows file server      Rehost        Quick win; no code change
VMware cluster           Relocate      Keep tooling, move the whole environment
Self-hosted email        Repurchase    Moves to a SaaS email product
On-prem MySQL app        Replatform    Same app, managed database (RDS)
Monolith order system    Refactor      Needs elasticity; split into services

DataSync task (CLI sketch)

aws datasync create-task \
  --source-location-arn arn:aws:datasync:eu-west-1:111122223333:location/loc-source \
  --destination-location-arn arn:aws:datasync:eu-west-1:111122223333:location/loc-dest \
  --name nightly-archive
aws datasync start-task-execution --task-arn arn:aws:datasync:eu-west-1:111122223333:task/task-0abc1234

The locations (source NFS/SMB share, destination S3 bucket) are created first and need an agent on premises.

Which AI service?

Goal                                             Service
-----------------------------------------------  --------------
Train and host my own model                      SageMaker
Add generative AI via foundation models          Bedrock
Build a voice or text chatbot                    Lex
Turn call recordings into text                   Transcribe
Pull fields from scanned invoices                Textract
Find sentiment or entities in reviews            Comprehend
Detect objects in photos                         Rekognition

How it works

A migration usually runs in phases: assess and discover the estate, plan with the CAF and the 7 Rs, build the landing zone (accounts, network, security), migrate in waves with replication tools, then optimise. Data transfer chooses between network (DataSync, VPN or Direct Connect) and physical devices depending on volume and available bandwidth - when the calculation says weeks or months online, shipping a device wins.

AI services come in layers: ready-made APIs (Transcribe, Textract), a platform to build your own models (SageMaker), and foundation-model access (Bedrock). Choose the highest layer that meets your need.

  Discover -> Plan (CAF, 7 Rs) -> Build landing zone -> Migrate -> Optimise
                                           |
                       Data: DataSync / network OR Snow device
                       Servers: Application Migration Service
                       Databases: SCT + DMS

Why does it exist?

Moving hundreds of applications without a plan creates chaos. Frameworks give a shared vocabulary for business and technical teams, while tooling reduces downtime and manual error. AI services let teams add intelligence without hiring a research lab.

When to use it

Use the 7 Rs when triaging an application portfolio, physical Snow devices for huge datasets over slow links, DataSync for recurring online transfers, and managed AI services before building custom models.

When not to use it

Do not refactor everything - that is the costliest strategy. Do not ship a device for a small dataset that a network transfer can handle. Do not train a custom model when a managed API already solves the task.

Common mistakes

  • Treating every app as a rehost and getting no cloud benefit.

  • Skipping the Retire step and migrating applications nobody uses.

  • Ignoring data gravity: apps move easily, databases and data do not.

  • Forgetting security and governance in the migration plan.

  • Mixing up Textract (documents), Transcribe (audio) and Comprehend (text analysis).

  • Starting a migration without discovery - hidden dependencies cause outages.

Practice exercises

  1. Easy:

    List the six CAF perspectives and group them into business and technical capabilities.

  2. Medium:

    Assign a Strategy to five applications of your choice and defend each.

  3. Medium:

    Compute how long 200 TB takes over a 1 Gbps link (use the unit converter) and decide whether to ship a device.

  4. Hard:

    Design a Textract plus Comprehend pipeline using S3, Lambda and SQS to process incoming invoices.

Interview questions

Name the 7 Rs.

Retire, Retain, Rehost, Relocate, Repurchase, Replatform, Refactor.

CAF perspectives?

Business, People, Governance, Platform, Security, Operations.

When would you use a Snowball Edge?

To move large datasets or process data at the edge when network transfer is too slow, expensive or unavailable.

Exam-style: Which service lets you build generative AI apps using foundation models through an API?

Amazon Bedrock.

Exam-style: Which service extracts text and tables from scanned documents?

Amazon Textract.

Exam-style: Which service collects on-premises server utilisation and dependency data to plan a migration?

AWS Application Discovery Service.

Exam-style: Which service provides a single dashboard to track migrations?

AWS Migration Hub.

Name the six CAF perspectives.

Business, People, Governance, Platform, Security, Operations.