IBM Support

EDCICONV performance considerations

Question & Answer


Question

Could output dataset's DCB attributes impact EDCICONV performance ?

Cause

Not choosing the optimal LRECL of output dataset could lead to additional overhead.

Answer

When converting data using EDCICONV, performance may vary depending on the LRECL of the output dataset.

The book states "Records from the input data set are processed by the iconv utility sequentially, one record at a time. If DBCS or multibyte character codes span a record boundary, the iconv utility will not recognize the partial sequence as a valid character and will fail. Therefore, input records must terminate with a complete character sequence. Be sure that the record length of the input data set is large enough to contain the longest input record."

For output dataset, it states "If the output data set has variable length records, the record length must be large enough for the longest record created."

Do note for a VB dataset, each record may be a different length. So in order to keep track of the record length and blocking, for every block, an additional 8 bytes are needed. The first 4 bytes of each block are reserved for the Block Descriptor Word (BDW); additionally the first 4 bytes of each record are reserved for the Record Descriptor Word (RDW)

For best performance, the LRECL of the output dataset should be determined by taking into account the length of the longest record that it is expected to be created after conversion and the additional 8 bytes needed to maintain the record information in VB format as data split over multiple records could potentially lead to slower performance.

[{"Line of Business":{"code":"LOB56","label":"Z HW"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SWG90","label":"z\/OS"},"ARM Category":[{"code":"","label":""}],"Platform":[{"code":"PF035","label":"z\/OS"}],"Version":"1.9"}]

Document Information

Modified date:
03 September 2021

UID

isg3T1012548