Serialize To File
Description
The Serialize To File transform stores rows of data in a binary format in a file. The binary format has an advantage over a text (flat) file because the content does not have to be parsed when read back. The metadata is stored in the file as well.
The Serialize to file transform supports a write-once access pattern, and does not support appending to existing files.
Options
| Option | Description |
|---|---|
Transform name |
Name of the transform. This name has to be unique in a single pipeline. |
Filename |
Name of the cube file to write. The path can contain variables. |
Include transform nr in filename |
Gives each transform copy its own file. The copy number is inserted before the extension, starting at 0. With this option, |
Create parent folders |
Create the parent folders when they do not exist. |
Do not create file at start |
Wait for the first row before creating the file. Otherwise the file is created when the pipeline starts, including the layout of an empty stream. |
Add filenames to result |
Add the written filename to the list of result filenames. |
Compression level |
How hard to compress the file: 1 - Fastest, 2 - Very fast, 3 - Fast, 4 - Balanced, faster, 5 - Balanced, 6 - Standard (the Java default), 7 - Smaller files, 8 - Very small files, 9 - Smallest files. A low level writes much faster and gives a somewhat larger file. De-serialize from file reads every level. A new transform uses 6 - Standard, and so does a transform saved before this option existed. |
Several files at once
Set the number of copies on the transform and enable Include transform nr in filename. Each copy writes the rows it receives to its own file. De-serialize from file reads those files back with the same option and the same number of copies.
A single copy writes one file. Several copies that share one filename overwrite that file.