Please enable JavaScript.

Coggle requires JavaScript to display documents.

Databases (RDS (OLTP) (read replicas (available for (Aurora (Aurora…

- - - - highly available
        
        2 copies in each AZ
        minimum 3 AZ
        making 6 copies
        
        designed to handle
        loss of copies
        
        self-healing storage
      - MySQL and Postgres
        compatible
        
        with much better
        performance
        
        much cheaper
      - starts with 10GB instance
        
        scales in 10GB increments
        to 64TB
      - compute scales up
        to 32vCPU and 244GB memory
      - backups
        
        automated
        
        always enabled
        
        do not impact performance
        
        snapshots
      - nodes
        
        writer
        
        reader
  - - - planned database maintenance
      - DB instance failure
      - AZ failure
    - - SQL Server
      - Oracle
      - MySQL
      - PostgreSQL
      - MariaDB
  - - - breaks replication
    - - MySQL
      - PostgreSQL
      - MariaDB
      - Oracle
      - Aurora
        
        Aurora replicas - 15
        
        can automate failover
        
        MySQL replicas - 5
        
        both asynchronous
  - - - recover to a point in time
        (down to a second)
      - retention period
        (1 - 35 days)
      - automatic,
        full daily snapshot
        including transaction logs
        
        done within a defined time window
        
        affects I/O
      - enabled by default
      - free, equal to the size of DB
      - deleted when RDS instance is
    - - manual
      - retained after RDS instance deletion
    - - new RDS instance
      - new DNS endpoint
  - - - uses the
        AWS Key Management Service
        (KMS)
      - underlying storage
      - automated backups
      - read replicas
      - snapshots
- - - - single node
        160GB
      - multi-node
        
        leader node
        
        client connections
        
        receives queries
        
        compute node
        
        store
        
        perform queries
        
        max. 128
        
        scale out by adding
        compute nodes
        
        priced by compute
        node hours
      - only available in 1 AZ
    - - automatic compression
    - - enabled by default
        
        1 day
        retention period
        
        max. 35 days
      - 3 copies
        
        original
        
        replica node
        
        S3
        
        can replicate
        to another region
        (DR)
    - - SSL in transit
      - AES-256 at rest
      - Redshift manages keys
        
        can use
        
        HSM
        
        AWS KMS
- - - - key/value pairs
  - - - spread across
        3 geographically
        distinct data centres
    - - SSD storage
    - - document
      - key/value
    - - eventually consistent reads
        
        data consistent
        after 1 second
      - strongly consistent reads
        
        can be configured
        to provide quicker
        data consistency
- - - - simpler
    - - multi-AZ
      - pub/sub
      - sorting datasets
      - backup/restore
      - persistence