HP Neoview Release 2.5 Software - Manual (Page 21)

• You cannot perform recovery if a job failed during transfer from the staging table to the
target table row, you must restart the entire operation.
• If recovery is not performed, the staging table is not deleted. Check the daily log file in
NVTHOME/log/java
to find the filename and drop the leftover staging table. The staging
table is deleted automatically if recovery is performed. For information about recovery, see
“Job Recovery” (page 80) .
Parallel Streams
Transporter supports the use of parallel streams for data movement. Whenever possible,
Transporter uses the parallel stream feature so that data movement is fast and efficient.
You determine the number of parallel streams with the “parallelstreams” (page 62) option in
the control file. For an extract operation, you also supply a filename prefix for the target filenames,
and Transporter creates the necessary target files.
Using parallel streams can increase performance for load and extract jobs. However, because a
larger number of parallel streams increases overhead on the client system, you must determine
the best number of streams to use for maximum performance.
TIP: Begin by using the default number of parallel streams and then tune your number of
parallel streams based on the performance you observe.
Parallel Query Extract
Transporter enables you to use a single source query to generate multiple parallel data streams
from the Neoview platform. For more information, see the
“parallelstreams” (page 62) control
file option.
Prehashing
Transporter streamlines the process for load operations with hash-partitioned tables by performing
a pre-hashing operation on the client. For more information about hash-partitioned tables, see
the HP Neoview SQL Reference Manual
NOTE: Pre-hasing cannot be performed for loads if the partitioning key cannot be identified
by Transporter. Some data necessary for performing the hashing process is not available to
Transporter on the client. For example, Field Expression map entries must be computed on the
server side.
JMS Trickle Feed For Data Loads
Trickle feed is a technique that allows for continuous updates of the database as the data in the
source system changes. Trickle Feed differs in this way from the use of flat files and named pipes,
which are considered "batch" techniques.
Transporter supports the use of JMS Trickle Feed as a data source for load jobs.
Transporter Job Statistics on the Neoview Platform
Transporter metadata tables on the Neoview platform maintain information and statistics about
jobs and control files. For more information, see
“Job Statistics on the Neoview Platform ”
(page 81)
Reflexive Update
A reflexive update is an update in which an input value provided can be a numerical value that
is added to or subtracted from the current column value. Use the source and target field mappings
Parallel Streams 21