Thanks again, LanX.

"The old approach is fine as long as you only read from the data structure without changing it..."

OK, good.  I'll just be reading it, so I guess it will work for me.  

I've just ran this which confirms it should work for me:

perl -MData::Dump -e '@hash{(key1,key2)} = ({a=>1, b=>2}) x 2;dd \%has
+h;print "hash{key1}{a}=$hash{key1}{a}\nhash{key1}{b}=$hash{key1}{b}\n
+hash{key2}{a}=$hash{key2}{a}\nhash{key2}{b}=$hash{key2}{b}\n"'
do {
  my $a = { key1 => { a => 1, b => 2 }, key2 => 'fix' };
  $a->{key2} = $a->{key1};
  $a;
}
hash{key1}{a}=1
hash{key1}{b}=2
hash{key2}{a}=1
hash{key2}{b}=2
[download]

Thanks again for all your help.