Sunday, July 12, 2020

RDS - Relational Database Service


RDS - Relational Database Service
-         
https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Welcome.html
-         A fully managed AWS service that makes it easier to set up, operate, and scale a relational database in the AWS Cloud. It provides cost-efficient, resizable capacity for an industry-standard relational database and manages common database administration tasks.
-         Apps accessing DB can't have root access
-         AWS performs patching and upgrades, auto-backup, easy scaling
-         If Multi-AZ is selected, AWS can perform synchronous replication b/w the primary and the standby
-         If Multi-AZ 
is selected, Automatic failover is available as well
-         REGIONAL; Multi-AZ = same region
-         Weekly maintenance window – can control when; or AWS will select for you randomly, 30 min
-         AWS will NOT do: database schema, setting performance tuning
-         Supported: MS SQL Server, Oracle, PostgreSQL, MariaDB, AWS Aurora, MySQL
-         Can BYOL – your own license OR License Included (buy via AWS)
-         Can have up to 40 db's per account
          §  10 can be Oracle or MS SQL under License Included (license purchased from AWS)
          §  BYOL – all 40 whatever you want
-         RDS uses EBS volumes (not instance store):
          §  General purpose – use for db w moderate i/o
          §  Provisioned IOPS – for high performance OLTP workloads (txn)
          §  Magnetic – small db w/loads
          §  Max capacity of DB associated storage (EBS): MS SQL 4TB, all others 6TB
-         Multi-AZsynchronous stby, same region. You can't read/write to stdby.
          §  Can't dictate which AZ will host the stby – AWS will select. Can view it once launched.
          §  Failover – 1 to few mins.
          §  Recommended to do connection retries to account for failover time period
          §  Recommended to use Provisioned IOPs for Multi AZ – speed up replication
          §  Failover scenarios:
                  § Loss of primary AZ
                  § Loss of n/w connectivity to the primary
                  § Compute (EC2) or storage (EBS) failure on primary
                  § DB Primary is changed
                  § Patching the OS of the DB instance
                  § Manual failover – perform "Reboot with failover" on primary
          §  During failover CNAME (endpoint) of DB is updated to point to the standby IP
          §  Recommended not to point your app to the RTD IP, use Endpoint instead for smooth automatic failover
-         Can control which subnets the DB is in – use subnet groups. The group must have a subnet in each of the AZ's in scope. AWS will assign an IP
-         You manage your DB engine configuration by associating your DB instances with parameter groups. Amazon RDS defines parameter groups with default settings that apply to newly created DB instances. You can define your own parameter groups with customized settings. Then you can modify your DB instances to use your own parameter groups

No comments:

Post a Comment