Minimize disk I/O contention in Oracle

Of all the configurable components of a geodatabase, storage is perhaps the most frequently and extensively customized. Likewise, each database administrator (DBA) has a preferred method for organizing both the logical and physical storage structures of the Oracle database, based on research and techniques that have proven successful in the past.

You have broad flexibility to design a storage model for your geodatabase that fits the specific needs of your data, applications, and existing management policies. ArcSDE has few strict storage requirements. You may choose to deploy an entry-level computer with a single data disk and one tablespace for GIS data, or a high-end server with dozens of disk arrays and hundreds of Oracle files, each successfully supporting their intended environments. Fortunately, you can adapt both ArcSDE and Oracle to take advantage of whatever resources are available to run your geodatabase.

To minimize disk I/O contention for Oracle databases, you can position frequently accessed files on separate disks, when possible, and group on the same disks frequently accessed files with infrequently accessed files. To do this,

  1. Estimate the size of all the database components and determine their relative rates of access.
  2. Position the components given the amount of disk space available and the size and number of disk drives.

Diagramming the disk drives and labeling them with the components help keep track of the location of each component. Have the diagram handy when you create your database.

Some recommendations on how to avoid resource contention on an ArcSDE geodatabase stored in Oracle are listed below. For an explanation of the Oracle components discussed here, such as tablespaces and segments, consult your Oracle documentation.


8/19/2013