The funny part is: I was really involved in some LDAP systems some years ago and everyone believed that they are "optimized for read". Someone made a benchmark (sorry, long lost) and found out that this is not true: Available LDAP systems (not only open source) where extremely slow with a medium size dataset. He then started to implement an LDAP directory server from scratch and he got some decent results (
http://www.fefe.de/tinyldap/).
Still, LDAP is mostly marketing, and without the support of some email clients and management tools, LDAP would already have gone the way of all ad-hoc technical "inventions".