Novell’s CDE: Adapting for Survival
In a Twilight Zone episode based on the short story “Steel” by Richard Matheson, fighting between humans is banned, and prize fights are fought by androids. One manager’s robot (Maxo), an older-style android, can’t compete against his newer, stronger rival and breaks, due to lack of money to fix it, after many fights. The manager, played by Lee Marvin, dresses up like an android and wins enough money to repair and upgrade his android. The moral here could be a variation on the skateboard shibboleth: “Adapt or Die!”
What does battling Maxo have to do with Novell’s changes to its premier certification? Novell’s Certified Directory Engineer (CDE), is undergoing a large transformation by adapting to market forces. Attaining a CDE represents that the certificate holder has advanced directory knowledge, particularly concerning proper network design, problem avoidance and problem resolution. We can examine the CDE transformation in three broad areas. One adaptation is streamlining, as fewer courses are suggested for preparation. Another change is the content, including operating systems used in class and the test, as these are significantly updated. The last alteration is a reduction in the number of tests required. By including Linux and Windows Server 2000, Novell may earn enough market share to continue to upgrade and sell its products for the far future.
Another interpretation of the “Steel” theme is as it relates to certification testing. Novell was first with certification testing, but there is only so much you can measure with a traditional form-based test. Since wizards can be used to accomplish most administrative tasks, typical questions, such as multiple-choice, true-false, drag-and-drop and one-word or phrase “free form” answers do not measure much of the candidates’ knowledge. Novell has a patent on the technology to discern the state of a machine, so with current virtual technology, either VMWare or Virtual PC, you can map learning objectives to using, not learning, features. This practical approach has a two-fold advantage: Novell measures what the candidate does, not what he knows, and this also helps dissemble any “paper” certification charge levied at the certificate holder. In the future, we can expect improved self-assessment options, as well as Web-based tools that make use of this technology. Reading between the lines from what I was told at BrainShare, we can expect more practical exams from Novell, including a test for the Certified Linux Engineer (CLE) and possible licensing of technology to other providers.
Both new and old CDE preparation courses involve disparate operating systems, with the new coursework including Windows 2000 server and Red Hat Linux. Presently, two courses (#991, Advanced NDS Tools and Diagnostics, and #995, Directory Server Technologies) are suggested preparation beyond the Certified Novell Engineer (CNE) to prepare for three tests. Details on what the present practicum includes are available at www.novell.com/training/certinfo/cde/practicum.html.
The new requirements consist only of one test: a revised practicum, #050-684. Aside from real-world experience, the only recommended preparation for that test is one new course, #3007: eDirectory Tools and Diagnostics. Beware: This is an advanced, high-powered class. You won’t finish all the material in the allotted time. Depending on your skill level, you may require significant work outside the classroom. A beta of the new practicum, lasting three hours, was given at BrainShare 2003 for a reduced fee. The new practicum is slated to debut in September 2003, at a revised price of $195, only at select Pearson VUE centers, versus the present price of $395. The reason for the price difference? The old practicum made use of remote technology, connecting to real servers, while the new practicum is local, making use of three virtual servers and one workstation/server, plus the test engine, on one PC via VMWare. VMWare was my downfall, as I shall describe.
The new practicum consists of a scenario that you must accomplish for an enterprise with four visible servers, a number of partitions, with a large number of replicas, in one tree. You will have some time to become familiar with VMWare. Read the directions carefully. You must have all replicas in their original positions, the tree must be completely healthy, and you should not remove eDirectory from any server—so you can forget those kill switches! You may reboot a server, which can take up to 10 or 15 minutes, but this isn’t necessary and means you have erred. The given scenario involves something like fixing replication, moving a division, adding a user or users or changing their rights, and fixing the reason that some users could not always log in or could not always access needed resources. Of course, to directly attempt the scenario you are given would be foolhardy, as the tree is not healthy! There will be at least four things wrong with the tree. You will want to do a full eDirectory health check before you do any operation. I found it helpful to draw a replica placement grid, but had to use two sheets of paper in landscape mode to do so. Insist on enough paper and desk space at the testing center. After querying other BrainShare beta testers, I found that one of the first things you should check is time synchronization, as well as replication.
One of the test designers insisted that time synchronization was not an issue in the new practicum, but I found that notion to be pure fiction. There will certainly be replica and partition problems: no synchronization, no master replica, no writable replica and so on. Other possible problems include schema synchronization and communication failures between servers. Remember from the #991 or #3007 class that the best way to fix directory problems does not necessarily involve running an automated or even a manual directory database repair. Recall that one of the solutions to a lab problem in these courses involved knowing and applying the difference between a “receive all” and “send all”—doing a directory repair would not have fixed the problem.
Other tools and utilities that you will use extensively include iMonitor and DSTrace. I prefer the command-line DSTrace, but you could use the GUI tool. Thinking back to the exercises in #3007, as well as the class material, you realize that obituary processing also consumed much class time. I am guessing that a stuck obituary, possibly because of an improperly removed server, would be a favorite problem for a test designer to give. Resources available include LogicSource; the knowledge base is unavailable at present but may be included in future versions on a CD—searching will be slow. This is because giving live Web access during an exam would be a security breach—you could leave notes at a friend’s server! Organization and time management are critical, so prioritize tasks—what you have done, what you need to accomplish—and check time remaining every so often. Familiarity with VMWare helps. When you switch screens, you must click on the new screen to change focus. When you forget to do so, the screen prompt asks if you want to terminate. The one time I didn’t look at the screen, I accidentally prematurely ended the session. Novell may change this in the released versions of the practicum.
In course #3007, the lab exercises you complete are very similar to each of the problems or broken functions that you face in the test. The key difference is that you have multiple problems combined in the practicum. Failure to fix any of these usually will mean that you cannot complete one or more of the tasks given in the scenario. New to the #3007 course, as compared to its predecessor, is the Live Fire! series of exercises, done at the very end of class. Live Fire! exercises, sometimes combining two problems in each of seven exerc