Configuring Disk Space for Multiple Coservers

Configuring your disks is possibly the most important task for obtaining optimum performance with VLDBs. Disk I/O is the longest portion of the response time for an SQL operation that scans a large amount of data. Extended Parallel Server offers the advantage of parallel access to multiple disks spread across many coservers.

Your goal should be to make it easy for a DBA to administer a large database server and to ensure that tables can be fragmented appropriately across disks and coservers for fully parallel processing. To accommodate multiple coservers and multiple nodes, perform the following steps:

• Create the dbspaces that the coservers manage on as many physical disk drives as possible. This step maximizes parallel I/O access to multiple disks. To prevent I/O access problem, if possible make sure that each disk can be accessed by only one coserver.
• Create standard device names and chunk path names across all coservers. Although you are not required to create standard device names across all coservers, standard device names make managing the database server easier. You must use unique chunk path names for each disk on a node. Be careful not to duplicate chunk path names.