Possible consequences of character conversion

For best results, try to avoid character conversions whenever possible, because conversions can potentially slow performance and sometimes cause data loss. The best way to avoid conversions is to use the same CCSID for all of your data.

Character conversion is the process of converting data from one CCSID to another CCSID. This process can occur when data is transferred between a remote and local system or when data is manipulated within the local system. For more information, see Situations in which character conversion occurs.

The best character conversion is no conversion, because conversion always has a performance cost. The cost depends on the extent of the conversion. For example, if you have a Unicode table and select every row into an EBCDIC application, the performance cost is probably noticeable. However if you issue a SELECT MAX(xxxx) FROM on a Unicode table, and then convert the result to EBCDIC, you might not notice the performance cost.

The second best conversion is a lossless conversion. A lossless conversion is one in which all characters in the source CCSID exist in the target CCSID and thus, no character is lost. For example, a conversion from CCSID 37 to CCSID 500 is lossless, because they both include the same set of characters. The difference between these two CCSIDs is the placement of 7 characters. These seven characters have different code points in each of these CCSIDs. A conversion from CCSID 1208 (UTF-8) to CCSID 1200 (UTF-16) is also lossless, because they both include the same repertoire of characters.

If the conversion is not a lossless conversion, certain characters might be lost. ("Lost" means that these characters are replaced by substitution characters.) Thus, the integrity of your data can be compromised.