A hash of hashes is probably what you should be using. Maybe something like:

open (my $fd, '<','./dump2.txt') or die "Can't open dump2.txt";
while (<$fd>) {
        chomp;
        my @dump = split(/\t/);
        my $tbl_last = "FIL";
        my %data;
        foreach my $it (@dump) {
                if ($it =~ m/([A-Z]{3})\.(.+)/) {
                my $tbl = $1;
                my ($col, $val) = split('=',$2);
#               print "$it\t$tbl\t$col\t$val\n";
                $data{$tbl}->{$col} = $val;
        }
}
print Dumper(\%data);
[download]

You can learn more about such data structures by reading perldsc.