RDS

• Six available – MySQL, MSSQL, Oracle, Postgre, MariaDB, Aurora
  • You can have auto and manual snapshots
  • Manual snapshot can persist even after the RDS is terminated
  • RDS can be restored to the latest second using transaction log and automatic snapshots
  • From disaster recovery perspective implement backup processes that meets your
    • Recovery Point Objective (RPO) – how far back
    • Recovery Time Objective (RTO) – time to recover
• NoSQL – Dynamodb
  • Fully managed distributed database
  • Eventual read consistency – read may loose most recent write(less than 1 sec)
  • Strong read consistency –  read will accounts for all writes(delay is accounted)
  • Data stored on SSD maintains 3 copies across AZ
• Redshift – is Amazon columnar Data-warehouse
  • It can have upto 128 nodes
  • Columnar compression give better performance
  • One Master node manages traffic
  • Keeps 3 copies (the original and replica on the compute nodes and a backup on Amazon S3). Retention from 1 day to 35 days
• Read Replica – Asynchronized copy of the database
  • Read replica is for high demand, high availability
  • Not available for MS-SQL database
  • Can have upto 5 read replica, which can be cross region
  • Backup must be turned on for read replica
• Stand by – Is different than read replica.
  • It is synchronous backup of primary
  • During failover DNS automatically points to stand by
  • During backup primary’s performance is impacted as I/O is suspended. For MariaDB, MySQL, Oracle, and PostgreSQL, I/O activity is not suspended because the backup is taken from the standby
• Aurora
  • Backups 6 copies, 2 in each AZ. Hence it is available in regions with at least 3 AZ
  • MySQL can be promoted to Aurora
  • Can have upto 15 read replica

Home | IAM | S3 | CloudFront | SQS | EC2 | RDS | Route53