Grid Database

The grid uses a database to store all grid data, including but not limited to the following examples:

• Grid runtime binary files

• Grid applications and associated configuration files

• Grid topology and runtime metadata

The content of the grid database is generally considered to be static, that is, non-transactional. The most common trigger of an update to the grid database is when configuration changes are made. The following are some examples of such changes:

• Add, remove or change a host

• Add, remove, upgrade or re-configure an application

• User management changes

• Grid configuration changes including but not limited to bindings and properties

Given the content of the database, it is critical to the operation of the grid. Loss of the database or corruption to the data will result in the inability to operate the grid and therefore consideration must be given to the infrastructure, security, and maintenance of the database server.

It is highly recommended that the database server be prepared for production purposes in that some or all of the following principles are employed. Please note that these are only intended as guidelines to enforce the need to consider the requirements that the grid database be highly available:

• Disks are configured in a RAID array employing redundancy (for example, RAID 1, 5 or 10)

• Database server is clustered either active-active or active-passive to cope with single server failure

• Database server employs multiple network interfaces with a common address

• Regular point-in-time backups are taken and their ability to be restored is validated at each occasion