Directory Services as the Foundation of Organizational Infrastructures

by Tom Adelstein

If you have followed any of my last six installments about LDAP, then you know we've taken a technical approach to the subject. I wrote the majority of the material in this series as part of an O'Reilly book entitled "Linux System Administration" or simply LSA. You can find a write-up on the book at this link.

The material on LDAP did not make it into LSA for a number of reasons. First, O'Reilly already has an excellent book on LDAP written by Gerald Carter. A 600 page book on Linux system administration doesn't need a repeat of existing work. Secondly, I found open source LDAP a strange animal with which to deal.

And that animal conjures up some interesting observations.

Linux and the open source community has many great projects. However, those projects have limitations in the context of a boundary based enterprise infrastructure. While Linux, as an example, does a beautiful job as an application platform, it lacks some essential elements for creating an enterprise environment capable of serving the needs of organizations. I blame that on a lack of vision and resources by and in the community. I see that as a niche in the competitive landscape.

Perhaps we know what to do to create a secure, manageable environment using Linux but others with extensive resources need to take on that monumental set of tasks required to turn the vision into reality. This is where we can discuss the context of a directory (LDAP) based infrastructure further.

An enterprise infrastructure is a collection of user accounts, groups, computers and other resources that share a common security boundary. Within that boundary, architects and system engineers must establish an environment that insures security and provides a concept known as least privilege. Least privilege and least rights means that any user has only those rights required for him or her to do their jobs and no more.

Directory based management does not always provide a common security boundary or the ability to lock down least privilege within the boundary. I look at Novell as a company with the tools necessary to provides an entire infrastructure with a common security boundary. But, Open Enterprise Server (OES) and Open Workgroup Suite are products for knowledge workers does not an enterprise make.

We can also vary for a moment from the enterprise and go right down to the local Linux PC. We could say that the lack of lock down policies at the local level are missing too. Also, where in the collection of computers users and accounts that would make up common security boundary does an administrator have the ability to provide a hierarchical set of policies that can carry through and override local and/or workgroup policies.Without that structure you can have an organization running amuck.

Going forward in the LDAP series, I want to tease out a theoretical model out further. We can call it a blue print, if you would like. It's an idea, in my opinion, whose time has come.

In this blog entry, I just wanted to make you aware of the possibilities.