User Guide

11.0
- 10.0

Back Next

Tuning Conventional Path Unload Processing

If you use the conventional path unload method, you can configure parameters for parallel unload processing and fetching LOB data to improve performance.

Click the

Runtime Settings

tab >

Conventional Path

view.

The following image shows this view:

To configure parallel processing, complete the following fields:

Field	Description
Threads number for unload statements	The maximum number of tables or partitions from which data can be unloaded in parallel. The default value is 1.
Threads per segment for unload statements	The maximum number of threads that Fast Clone can use to unload data from a single table or partition. If you use multiple threads for a single table, CPU usage might increase. Usually, multiple threads are used to unload data from large tables. The default value is 1.
Use direct file access to fetch extent map	Select this option to have Fast Clone try to read extent information for parallel unload processing directly from a data file. By default, this option is selected.

With the conventional path unload method, Fast Clone uses one thread per core or CPU to unload data from an Oracle source. If the system has multiple CPUs, Fast Clone can perform parallel unload processing for the source tables.

To determine the total number of unload threads to use, multiply the number of tables from which data is unloaded in parallel by the number of threads for each table. For conventional path unload, the appropriate number of concurrent threads is equivalent to the number of cores or CPUs. If

I/O

bandwidth or the number of allocated CPUs is restricted, you might need to use lower the value.

If you need to unload data from many tables of similar size, set

Threads number for unload statements

to the appropriate concurrency and set

Threads per segment for unload statements

to 1.

If you need to unload data from a few large tables, increase the

Threads per segment for unload statements

value and set

Threads number for unload statements

to the number that you calculate as follows:

number of cores or CPUs ÷ number of threads per table

To configure settings for fetching LOB data, complete the fields described in the following table:

Field	Description
Fetch LOBs as single chunk	Select this option to unload a single chunk of LOB data that has the size specified in the Fetch LOBs as single chunk, in KB field. By default, this option is not selected and Fast Clone fetches all LOB data in multiple chunks.
Fetch chunk size for LOBs, in KB	To unload all of the LOB data in chunks of a consistent size, enter the size, in kilobytes, to use for each chunk. Default is 128 KB.
Fetch LOBs as a single chunk, in KB	If you selected Fetch LOBs as a single chunk , enter the maximum size, in kilobytes, of the single chunk.
Fetch array size	The number of rows in an array that Fast Clone unloads from the database during each round trip for conventional path unload processing. Default is 500.

Rename Saved Search

Table of Contents

User Guide

User Guide

Tuning Conventional Path Unload Processing

Tuning Conventional Path Unload Processing