The Integration service supports two collations: Unicode binary and ASCII case-insensitive. The default collation is Unicode binary. The Integration service collations are explained in Table 13-1.
|
Table 13-1
|
Integration service collation
Unicode code point order (binary order). All characters are different from one another and are sorted by their Unicode values.
'E' < 'e', since E = U+0045 and e = U+0065.
'o' < 'Ö' (O with an umlaut), since o = U+006F and Ö = U+00D6.
ASCII code point order, with uppercase characters given the same value as lowercase characters.
'E' = 'e' since case is not considered.
Cannot sort Ö, since it is outside the ASCII range (U+00D6). ASCII range is between U+0000 and U+007F.
'E' > '[', since E = U+0065 and
[ = U+005D.
[ = U+005D.
Choose ASCII_CI if your databases contain only ASCII characters and case-insensitive sorting is required.
When the Integration service collation matches the database collation, all string comparison and sort operations are sent to the database. When the Integration service collation does not match the database collation, some or all string comparison and sort operations must be performed by the Integration service. For this reason, performance is optimized when the Integration service collation matches the collation on as many of your databases as possible. For example, you have five different databases, all of which contain only ASCII data, and either case-insensitive or case-sensitive sorting is acceptable. You can choose either ASCII_CI or UNICODE_BIN as the Integration service collation. However, if four of the databases sort case-insensitively, while the fifth database sorts case-sensitively, you should choose ASCII_CI so that performance suffers only when the Integration service must compare and sort strings from the fifth database.
