Solving Operational Business Intelligence with InfoSphere Warehouse Advanced Edition

Chapter 8. Building a corporate backup and recovery strategy 301

tables to be unloaded from in parallel where you want to reduce the resources

consumed.

򐂰 Controlling processor resources with the nbcpu configuration file parameter

The nbcpu configuration parameter specifies the maximum number of

threads allowed when unloading data. Default behavior is to start a work unit

for each processor core. This parameter is set in the db2hpu.cfg configuration

file and cannot be overridden in the control file.

In a scenario involving a database outage, you might want to allow Optim

High Performance Unload to use all available processor resources to unload

data as quickly as possible. In an environment where queries are running

during recovery, configure Optim High Performance Unload to restrict the

resources consumed, thus reducing the amount of processor used.

8.4.4 Optim High Performance Unload with named pipes or output

files

When using Optim High Performance Unload, you can choose to unload data to

output files for subsequent processing, or you can unload data directly to named

pipes for immediate processing by another utility or application such as DB2

LOAD or DB2 CDI.

Optim High Performance Unload uses a staging area to stage data pages read

from the backup image. Data is then written to output files in the output area or to

named pipes:

򐂰 Output files: Unloading data to output files allows this process to be separated

from the process of loading data, allowing for validation or postponement of

the load operation.

򐂰 Named pipes: Using named pipes allows data to be unloaded from a backup

image and loaded to the target database table concurrently, and uses less

disk space to perform the recovery process.

To facilitate parallel unload and load operations through named pipes, create a

control file to unload data for each database partition to individual named pipe file

markers on the appropriate node. Then use the DB2 Load utility to load data in

parallel across all the database partitions, thereby avoiding unnecessary network

traffic between data nodes during the recovery process. The syntax or use of the

DB2 Load utility is not altered by Optim High Performance Unload. All features

and parameters of the DB2 Load utility are supported and operate normally.

When using the DB2 Load utility to complete a recovery scenario, we

recommend that you perform a backup of the recovered data in line with your

service level objectives for backup and recovery.