AWS Cloud Practitioner #24: Migration y Innovation Services

¿Qué es?
Move application AS-IS a AWS (no changes)
 
Ejemplo:
VMware VM → EC2 instance (same OS, same app)
 
Pros:
✅ Rápido (weeks vs. months)
✅ Low risk
✅ No re-architecture needed
 
Cons:
❌ NO cloud-optimized
❌ Miss cloud benefits
 
Cuándo:
- Quick migration needed
- Legacy apps (can't modify)
- Initial step (rehost → optimize later)
 
Tools:
- AWS Application Migration Service
- CloudEndure Migration

¿Qué es?
Minor optimizations durante migration
 
Ejemplo:
On-premise MySQL → Amazon RDS
(Same database, pero managed)
 
Pros:
✅ Some cloud benefits (managed service)
✅ Minimal changes
✅ Faster than refactor
 
Cons:
❌ Not fully cloud-native
 
Cuándo:
- Want some optimization
- Can make small changes
- Balance speed vs. benefits
 
Example:
App server → EC2 (rehost)
Database → RDS (replatform, managed)
File storage → S3 (replatform)

¿Qué es?
Move to different product (usually SaaS)
 
Ejemplo:
On-premise Exchange → Microsoft 365
Custom CRM → Salesforce
 
Pros:
✅ Modern features
✅ No maintenance
✅ Scalable
 
Cons:
❌ Vendor lock-in
❌ Migration effort (data, training)
❌ Recurring cost
 
Cuándo:
- SaaS alternative exists
- Don't want to maintain
- Want latest features

¿Qué es?
Redesign application para cloud-native
 
Ejemplo:
Monolith → Microservices
EC2 → Lambda + API Gateway + DynamoDB
 
Pros:
✅ Max cloud benefits
✅ Scalability, resilience
✅ Cost optimization
 
Cons:
❌ Time-consuming (months/years)
❌ Expensive
❌ Risk (major changes)
 
Cuándo:
- Need scalability
- Strong business case
- Technical debt opportunity
 
Example:
Old app: Single server, MySQL
New app: Lambda (serverless), DynamoDB (NoSQL), S3 (storage)

¿Qué es?
Identify and decommission unused apps
 
Example:
Audit: 100 applications
Used: 75
Unused: 25 → RETIRE
 
Pros:
✅ Reduce costs
✅ Less complexity
✅ Focus on important
 
Why migrate algo que no usas?

¿Qué es?
Keep on-premise (don't migrate yet)
 
Razones:
- Not ready (dependencies)
- Compliance (must be on-premise)
- Plan to retire soon
- Major refactor needed (not now)
 
Strategy:
Migrate 80% now
Retain 20% para later

Features:
✅ Discover applications
✅ Track migration progress
✅ Integrate con migration tools
✅ Visualize dependencies
 
Supports:
- Application Migration Service
- Database Migration Service
- CloudEndure Migration
- Server Migration Service
 
Dashboard shows:
- Applications discovered: 150
- In progress: 45
- Completed: 30
- Not started: 75

Problem:
Need to transfer 100 TB a AWS
 
Over internet:
100 TB over 100 Mbps = 92 days
Cost: High data transfer fees
 
Snow device:
Ship device → Load data → Ship back → Upload to S3
Time: 1 week
Cost: Fixed device fee
 
Rule of thumb:
> 10 TB → Consider Snow
> 100 TB → Definitely Snow

Smallest:
- 8 TB storage
- 4 GB RAM
- 2 vCPUs
 
Use cases:
- Edge computing
- Remote locations
- IoT data collection
 
Size: Lunch box
Weight: 4.5 lbs

Two types:
 
Storage Optimized:
- 80 TB storage
- 40 vCPUs
- 80 GB RAM
 
Compute Optimized:
- 42 TB storage
- 52 vCPUs
- 208 GB RAM
- Optional GPU
 
Use cases:
- Large data migrations
- Local processing
- Temporary data center
 
Size: Suitcase
Weight: 50 lbs

Massive:
- 100 PB storage (!!!!)
- Exabyte-scale
- Shipping container on truck
 
Use cases:
- Data center closure
- Entire data center migration
 
Literally a truck arrives

1. Request device (AWS Console)
2. AWS ships device to you
3. Connect to your network
4. Copy data to device
5. Ship back to AWS
6. AWS uploads to S3
7. Device securely erased
 
Encryption:
All data encrypted (256-bit)
Keys managed via KMS
 
Tracking:
GPS tracking
Tamper-evident

¿Qué es?
Migrate databases a AWS
 
Features:
✅ Homogeneous (Oracle → Oracle on RDS)
✅ Heterogeneous (Oracle → PostgreSQL)
✅ Minimal downtime (continuous replication)
 
Process:
1. Create replication instance
2. Define source (on-premise DB)
3. Define target (RDS)
4. Create task
5. Start migration
 
Source stays online durante migration
Continuous data replication
Cutover cuando listo
 
Supported:
Sources: Oracle, SQL Server, MySQL, PostgreSQL, MongoDB, etc.
Targets: RDS, Aurora, Redshift, DynamoDB, S3

¿Qué es?
Platform para build, train, deploy ML models
 
Use case:
Data scientist trains model
Deploys a production
Realtime predictions
 
No need:
- Setup infrastructure
- Manage servers

Image/Video analysis:
 
Features:
- Object detection
- Face recognition
- Text in images
- Content moderation
 
Example:
Upload foto → Rekognition → "Person, Dog, Beach"

Natural Language Processing (NLP):
 
Features:
- Sentiment analysis (positive/negative)
- Entity extraction (names, dates, places)
- Language detection
 
Example:
"Amazon is great!" → Sentiment: Positive (0.98)

Text-to-Speech:
 
Input: "Hello world"
Output: Audio file (MP3)
 
Languages: 60+
Voices: Natural-sounding
 
Use case: Accessibility, audiobooks

Speech-to-Text:
 
Input: Audio file
Output: Transcript text
 
Features:
- Automatic punctuation
- Speaker identification
- Custom vocabulary

Language translation:
 
Input: "Hello" (English)
Output: "Hola" (Spanish)
 
Languages: 75+
Neural machine translation

Connect IoT devices a cloud:
 
Devices:
- Sensors
- Smart home devices
- Industrial equipment
 
Features:
- Device management
- Secure communication (MQTT)
- Rules engine (trigger actions)
 
Example:
Temperature sensor → IoT Core → Lambda → Alert si > 80°F

Query S3 data con SQL:
 
Data: Logs en S3
Query: SELECT * FROM logs WHERE status=500
Result: Table
 
No servers, pay per query

Business Intelligence (BI):
 
Data sources: S3, RDS, Redshift, etc.
Output: Interactive dashboards
 
Visualizations:
- Graphs, charts, tables
- Drill-down
- Mobile-friendly

Real-time data streaming:
 
Use cases:
- Log analytics
- IoT telemetry
- Clickstream data
 
Components:
- Kinesis Data Streams (collect)
- Kinesis Firehose (load to S3, Redshift)
- Kinesis Analytics (SQL queries on streams)