LDAP Series Part V - Getting a Grip on Directory Service Modeling

by Tom Adelstein

I have an observation I'd like to disclose about the Open Source community: We tend to leap into all kinds of things before we have all the facts and/or information necessary to make intelligent decisions. We criticize other communities, laugh at things like directory services from the two major NOS players, talk about all our great applications, etc. We hang on to old notions about what makes Linux tick. Sorry, but that model ESR defined doesn't fit any more. The community natter appears to come mostly from people who lack deep technical skills and knowledge of enterprises.

While Linux has garnered a major part of the UNIX market, it has not made much progress in the enterprise management field. Without directory services to create a serious model of an enterprise, Linux will continue to remain a great application server. Under Novell, Linux will become a nice kernel for the Netware proprietary stack.

I'm also concerned about the technology leader, Red Hat. Their inability to utilize the assets purchased from AOL demonstrates a lack of vision. With Directory and Certificate servers, Red Hat has the ability to provide Identify Management, user management and a more secure network environment. It needs to move quickly because its competitor, Novell, has Open Enterprise Server and that puppy provides outstanding enterprise tools.

Where to Start

An LDAP directory service provides the framework for enterprise management. Open Source LDAP servers need numerous features to compete and evolve into an identity management system. Running OpenLDAP or Fedora Directory Server from the command line may work for some but without a visual model and the ability to replicate across an enterprise transparently OSS LDAP stagnates. Also, the lack of a visual tool keeps OSS advocates from learning how to use OpenLDAP as an enterprise directory. FDS has a visual interface that's outdated and doesn't provide features useful across the enterprise.

Learning OpenLDAP and/or FDS starts with what seems like unnecessary root level orientation. The model focuses on setting up the top of the tree. That may appear like a place to start if you’re a complete geek who loves to fool around the hacking hardware code. It doesn't do much for an administrator.

Admins need the ability to focus on Organization Units (the ou) and model their organizations in the directory sever. We need to manage departments, people and resources across an enterprise. I want to see a set of organizational units under the auspices of a root server and I want to manage my mail, dns, dhcp, web services, shares, users and security. But unless you have lots o' bucks for Novell, the typical admin cannot do that.

An emerging OSS Organizational Model?

Unfortunately and maybe fortunately, Novell needs a low cost competitor. I suspect that such a development group will emerge as a startup. I'd like to see such an effort come from the Debian community. It's even OK with me if the Ubuntu team puts it together. I believe the effort will require a large team of dedicated developers who can finish a project.

I don't expect Red Hat to do this. Red Hat is already stretched thin meeting its low cost business model. Additionally, for perhaps the first time, Red Hat may have problems competing with Novell. As a side note, I can see the latter going after the best people at Red Hat as long as Novell does a Chris Stone with their monkey managers. I wouldn't work for either of those chimps.

Also expect Redmond's Open Source Software Lab to work with Novell to allow it into the forest. Redmond lacks some serious management tools. For example, have you ever attempted to run any command to see who is logged on to a server in a MS enterprise? Run any command you wish and you won't see what we can do with a simple command like “who