How large is large?
Personally I tend to use Data::Dumper. (I prefer setting $Data::Dumper::Indent to 1.) Admittedly that is slow and memory wasteful for data structures with tens of thousands of elements.
Some people prefer the output of YAML. I haven't used it though.
Either will dump a hash of arrays in a readable format, and won't break if your data structure isn't quite as regular as you thought.
Beyond that, if you are going to write your own code for dumping a hash of arrays, I would strongly recommend indenting the array elements some and outdenting the hash keys so that visual scan can tell you where you are in the data structure.
Cheers,
Ben
"good ideas and bad code build communities, the other three combinations do not"
- [link|http://archives.real-time.com/pipermail/cocoon-devel/2000-October/003023.html|Stefano Mazzocchi]