|By Ranko Mosic||
|January 28, 2013 03:34 PM EST||
Modern enterprise class Oracle shops typically operate hundreds of Oracle databases, all managed by sizable DBA teams.
Properly deployed Oracle databases actually require little administrative effort. I am here exploring ideas, technologies and techniques that bring us closer to the fully autonomic, self-managed Oracle database shops.
Autonomic computing aims at delivering self-managed and adaptable database systems ( and other IT components), while hiding complexity from users and operators. The goal is to minimize manual administrative effort, thus leading to reduced OPEX, better service and higher uptime.
Over the last few decades Oracle RDBMS has grown into mature, sophisticated product, featuring many improvements that bring it quite close to becoming Self-Managing Database System.
Database environments that closely adhere to an autonomic paradigm can be deployed today with the proper usage of Oracle database features, deployment techniques and specialized tools.
Database installations and configuration for most of the environments (i.e., non-production) should be done in a private cloud environment based on templates. Fully configured database, including backups, monitoring etc. can be deployed in less than an hour. An example of a public cloud environment that can be emulated in a private cloud setting are Amazon Web Services IaaS and PaaS platforms.
An interim solution could be that database installations are performed using response files, templates or gold images.
Oracle RDBMS optimizer produces best SQL query execution plan based on collected statistics, almost eliminating the need to manually tune SQL statements.
Database disk space and memory management are fully automated. ASM (Automated Storage Management), datafile autoextend, ASSM (Automated Segment Space Management) and AMM (Automatic Memory Management) handle all disk space and memory management tasks.
Oracle Instance Caging (database instance throttling) and Database Resource Manager (Exadata also features IORM) make it possible to divide and prioritize available resources between database instances and session groups, thus limiting their ability to overburden and destabilize the system.
Oracle Restart (for single instance databases), or Oracle Clusterware (in RAC environments) can automatically restart components like listener, database instance etc. after component failure or node restart. Multi-tenancy (also coming as an explicit feature in the upcoming Oracle 12c database) make it possible to reduce administration burden by sharing resources between multiple tenant applications.
Corrective actions can be specified in Oracle Enterprise Manager (SQL, host scripts to be executed) after certain predefined condition is met i.e. in response to policy violations.