In this post (http://www.coactivate.org/projects/topp-engineering/blog/2009/03/20/to-zodb-or-not-to-zodb) the author presents two characteristics (disadvantages?) of ZODB:
- No ad-hoc queries.
- Migrations are inconvenient and expensive.
ZODB is very interesting but, IMHO, the major drawback is that the data is dependent of original code.
Grok is a good framework, but maybe you can look Pyramid (www.pylonsproject.org), that has also support for traversal and ZODB, with the advantage of a more numerous community.