It seems that the first problem is to match the keys in the seperate hashes and then compare the contents so something like this might work (untested)

foreach my $sqlkey (keys %sql){
    foreach my $csvkey (keys %csv){
        next unless $csvkey =~ /$sqlkey\z/;  # match end of sqlkey
        # code to compare $csv{$csvkey} & $sql{$sqlkey} and do other t
+hings
    }
}
[download]

Of course, this is an O(m*n) operation for m keys in %sql and n keys in %csv. If %sql and %csv are very large, this might be a prohibative operation. If this is true, you might want to map the keys of %csv to the keys of %sql (assuming they are one to one) and then do a straight comparison.

   my %sqltocsv = qw(EMP_NAME NAME EMP_JOB JOB ...);

   foreach $sqlkey (keys %sql){
       if($sql{$sqlkey} eq $csv{$sqltocsv{$sqlkey}}){
           # do things
       }
    }
[download]