Before you can search BLOB data, you must prepare a memory area for storing BLOB data in the client. You will also need a send buffer for BLOB data returns in the server and memory for a receive buffer that accepts BLOB data in the client library. Consequently, a large amount of memory must be allocated according to the BLOB data size, and the memory resources will be strained.

More and more systems are configured so that a middleware program that operates as a HiRDB client is placed between the end user program and HiRDB. This configuration design has further increased the amount of memory used as BLOB data is transferred between these programs.

The file output facility for BLOB data prevents increased memory usage during BLOB data searches by outputting retrieved BLOB data directly to a file in a single server or a unit with a front-end server, instead of returning the BLOB data to the client. The facility then returns the name of the file to the client.

Figure 4-65 shows an overview of the file output facility for BLOB data.

Figure 4-65 Overview of the file output facility for BLOB data

Explanation

1. When the client searches BLOB data, the server outputs that BLOB data in single rows and single columns to a file.
2. The server returns the file name of the BLOB data that was output in (1) to the client.
3. Based on the file name that was returned, the client accesses the BLOB data file located in the server.