Well, it depends. In your second code example, that makes it look as if you're calling my_function on your $DBH object. You can't do this unless you've defined my_function to be in the class into which your database handle is blessed (or in some ancestor of that class). In your case, that ancestor would be DBI. I'm not sure that inheriting from DBI would be particularly easy, considering that it relies on specific database drivers to do much of the work. So in that case, what exactly are you inheriting from: DBI or the database driver?

Here's what I've done in the past, which makes it easier. I use a "has-a" relationship rather than an "is-a" relationship.

I've created a My::DBI class (or something such). An object of type My::DBI contains a DBI database handle; but it's not actually a database handle itself. Every method that I call on my database handle is forwarded to a call on the database handle; before and after this forwarding I can do some processing on my own. For example (this code is untested, though it's based from memory on code I've written before, which *is* tested):

    package My::DBI;
    use strict;

    sub new {
        my $class = shift;
        bless { @_ }, $class;
    }

    sub connect {
        my $self = shift;
        return $self->{dbh} if defined $self->{dbh};
        croak "Need connection parameters"
            unless defined $self->{connect};

        my $c = $self->{connect};   ## Simplify typing :)
        return $self->{dbh} = DBI->connect
            ($c->{dsn}, $c->{user}, $c->{pass}, 'mysql',
             { RaiseError => 1 });
    }

    sub my_function {
        my $self = shift;
        ## $DBH is now in $self->{dbh}
    }

    ...
[download]

And so on. Usage:

    my $db = My::DBI->new(connect => { ... });
    $db->my_function(@args);
[download]

Would that work for you?