Question

What is internal working of db2 for comparing strings ?

i have a query

select * from mytab where name=’rohit’

how name in each row is compared to value ‘rohit’

I am also thinking to have a column having hash code for each name(varchar) column having index on it. so whenever i look for a name, i should be looking for hashcode instead.

is this approach feasible and can gain performance?

Answer 1

First Question

What is the internal working of DB2 for comparing strings ?

The simple answer is case sensitive matching.

The more complicated answer is provided by IBM. IBM is famous for complicated answers.

Two strings are compared by comparing the corresponding bytes of each string. If the strings do not have the same length, the comparison is made with a temporary copy of the shorter string that has been padded on the right with blanks so that it has the same length as the other string.

Two strings are equal if they are both empty or if all corresponding bytes are equal. An empty string is equal to a blank string. If two strings are not equal, their relationship (that is, which has the greater value) is determined by the comparison of the first pair of unequal bytes from the left end of the strings. This comparison is made according to the collating sequence associated with the encoding scheme of the data. For ASCII data, characters A through Z (both upper and lowercase) have a greater value than characters 0 through 9. For EBCDIC data, characters A through Z (both upper and lowercase) have a lesser value than characters 0 through 9.

Varying-length strings with different lengths are equal if they differ only in the number of trailing blanks. In operations that select one value from a collection of such values, the value selected is arbitrary. The operations that can involve such an arbitrary selection are DISTINCT, MAX, MIN, and references to a grouping column. See the description of GROUP BY for further information about the arbitrary selection involved in references to a grouping column.

Second Question

I am also thinking to have a column having hash code for each name(varchar) column having index on it. so whenever I look for a name, I should be looking for hashcode instead.

Is this approach feasible and can gain performance?

Yes, you can hash names. No, you won’t see any performance improvement.

The reason is that most of the time that DB2 spends going through a table or table index is used in disk input. The string matching of a name index is a tiny part of the processing time.

One thing you might consider implementing is a soundex matching of names.