The most basic connection is from a dataset to a physical file.

There are two processes for connecting to physical files:

This process can be used to open, read, and parse data from any fixed record length sequential file, such as COBOL legacy data. This includes the ability to read ASCII or EBCDIC and text or binary data of virtually any type or style (for example, COBOL packed, reverse byte-order, old floating point formats, and blobs). You can define rules governing a particular flat file and its structure and then extract clean data records for transformation.

Because of its binary reading capability, the integration platform can extract data from unknown raw file formats. Most commercial applications store data records using fixed-length techniques. When this is the case, the integration platform can be used as a brute force extractor to open the file, navigate to the beginning of fixed data, and rip records out according to rules that you create.

For related information, see Binary.

This process is used to physically read and write in the native internal storage format of the file. For example, you can use this process with xBASE files. The integration platform can use this process when there is published information describing in detail the internal storage structure of the file.

The advantage of this process is that it provides fast access, high performance, and code that can be used across platforms. However, because it requires developers to keep up with changes to file formats, this process is not often used. It is, however, an option for certain file formats, particularly text formats and those that are old or obsolete.