Databases on AWS

Question 1

Databases RDS

Answer 1

Relational Databases (RDS)

• Multi-az - for desaster recovery - From an EC2, if primary DB goes down it flips over to a secondary DB in new AZ
• Read Replicas - Whenever write happens in the primary DB it copies over to a read replica which you’d then have to link up with your EC2 separately

Types

• MySQL
• Oracle
• SQL Server
• Auroa
• MariaDB

Non relational Databases

• Can change/add a key:value pair to an object without adding them to all objects

Test tips

• RDS runs on virtual machines
• RDS is not serverless - Aurora is
• Cannot log into these OSs
• Patching RDS OSs and DBs is amazons responsiblity

Question 2

Backups, multi-AZ, read replicas

Answer 2

Backups

• Automated backups are stored in S3
• SNAPSHOTS are done manually
• Whenever your restore a backup or snapshot it will have a new RDS

Multi-AZ

• Allows for an exact copy of your database in another AZ
• For disaster recovery only
• Can force a failover

Read Replica

• Allows for a read only copy of your database
• Async/Multi-AZ
• Used for scaling
• Up to 5 copies
• Will have it’s own DNS endpoint
• Can be promoted to it’s own database

Question 3

DynamoDB

Answer 3

DynamoDB

• NoSQL database service

Basic info

• Stored on SSD Storage
• spread across 3 geographically distinct data centers
• Eventual Consistant Reads (default)
  
  • Consistancy across all copies of data around 1 sec
• Strongly Consistant Reads
  
  • Less then 1 sec

Question 4

Redshift

Answer 4

Redshift - Petabyte scale data warehouse servuce in the cloud - used for business - one AZ currently

• Singel node (160gb)
• Always tries to maintain 3 copies of your data
• Multi-node
  
  • Leader node (manages client connections and receives queries)
  • Compute Node (store data preforms queries and computations)
    
    • Up to 128 compute nodes
• Highly advanced compression
• Massive parallel processing(MPP)
  
  • Using nodes
• Backups enabled by default - 1 day
  
  • retention period 35 days
• Can Async replicate to S3
• Priced by compute node hours
  
  • Backups as well

Question 5

Aurora

Answer 5

Amazon’s Aurora - MySQL and PostgreSQL-compatible relational database.

• 5 times better performance then MySQL, 3 times better than PostgreSQL at a much lower price point
• Serverless - On-demand for infrequent/intermittent or unpredictable workloads

Basics

• Starts at 10GB scales to 64GB
• Compute resources 32vCPU and 244GB of Memory
• 2 copies of your data in each AZ - min of 3 AZ == 6 copies
• You can share aurora snapshots with other AWS accounts
• 3 types of replicas avaliable
  
  • Aurora replicas(automated failover), MySQL, PostgreSQL
• Backups on by default

Question 6

Elasticache

Answer 6

Elasticache - web service that makes is easy to deploy, operate and scale an in-memory cache in the cloud. Stores things in a cache to make things load faster.

• Memcached
  
  • Simple
• Redis(Better)
  
  • Multi-AZ
  • Backups

Question 7

Database Migration Service (DMS)

Answer 7

Database Migration Service (DMS) - use to migrate databases on or off the cloud

• Hermogenous conversion - same DB engine type conversion
• Hertrogenous - one databases converts to a different type of database engine using Schema Conversion Tool (SCT)

Question 8

Caching Services

Answer 8

Caching Services

• CloudFront - media
• API Gateway
• ElastiCache - memcached and Redis
• DynamoDB accelorator

Question 9

EMR

Answer 9

EMR - Big data PROCESSING broken down into clusters

• Can confgiure clusters to periodically archive the log files stored in the master node in S3 five min intervals

EMR Cluster Node Types

• Master node - manages the cluser - status of tasks - log data stored here
• Core Node - software components - run tasks and stores data
• Task Node(optional) - does not store data - only runs tasks