SCEA Part I A survey licked off proceedings. It asked what your perceived ability across certain architectural areas was. I answered none to all just in case it affected the level of questions although I hoped it would be further from the truth than that! It then went on to an annoying disclaimer a couple of questions about wanting to receive information from Sun and their partner. This was bad enough, but to make matters worse, I noticed the clock had already started ticking down. The questions seemed to be batched by subject. The test included a lot of EJB. There was a reasonable amount but an easy level of UML � what methods were called by what in the sequence diagrams, what the relationship between classes and interfaces were in class diagrams. Patterns were covered as much as UML. There were no questions that asked you to name a pattern from UML despite the objectives. There were questions about data access objects and their relationships with EJB. There were explicit Internationalisation and Security questions as per the objectives but the summary at the end didn�t reflect this. It seemed like there were more radio questions that the previous SCJA and SCJP exams and a lot of the multiple answer questions specified the number of answers there should be. In some cases there were 7 options, which is more than I remember from the other exams. Thankfully there were no text entry based answers. (Tony Alicea could get 100% on this one!) I finished the first pass with 15 minutes to spare but had missed some questions intentionally as they were not straightforward. Brfore I started the test, I wrote a table of question numbers along with a space where I could grade how well I though I�d answered the question. This would enable me to concentrate my efforts on the ones I wasn�t sure off and forget the ones I was certain about. 48 questions within 75 minutes gave about 90 seconds for each question. The questions ranged from one-liner radio questions with a couple of options through to a page full of description and then multiple answers. A new concept of exhibits has been introduced which adds to the reading. The exhibits do provide a handy tile option although the resolution of the monitors that Interquad in Middlesex Road nr Liverpool Street, London made this ability relatively useless. Some of the longer questions took a good minute to digest, never mind answer. A lot of the questions touched on subjects that were certainly not covered in the objectives. Some of which I knew absolutely nothing about. One question expected you to know the port numbers that various protocols go through. I only got this question because the following question stated what went through the protocol I didn�t know so this may have been intentional. Firewalls are still in there with a vengeance. I wasn�t expecting this although I guess I should have been. It still covers HTTP tunnelling. A couple of questions touched on data access objects, which I hadn�t studied. I recently went through the J2EE sample test on the IBM ICE site. This was extremely irritating as did not distinguish between EJB 1.0 and 1.1. It was difficult to know which spec it was aimed at and this affected some of the answers. One in particular is the return types of certain EJB client API methods, you know the one, where it was void but now isn�t. How on earth do you answer this when it changes across the specifications? Unfortunately the Sun exam doesn�t specify this either which leaves you guessing a bit with some of the questions. Timeouts are now specified in a vendor specific manor not in the Deployment Descriptor. I can�t believe all those people who sat the beta missed this, and if they did point it out, why is it still in the exam. Sun have once again failed to produce an error free exam. There was an unexpected amount of references to XML despite it not being in the objectives explicitly. There seemed to be a lot of architectural design question where you are presented with a real-life design scenario and you have to make recommendations. This is cool as this is what you�re doing this for. It�s a shame you aren�t told which ones you get right and wrong. It could have a major impact on the systems you develop in the future! Anyway, I got 79%. It�s better than the 75% I got on the SCJA. As for preparation, I read the first 7 and a bit chapters of the Monson-Haefel book and I would strongly recommend this (although I would read more than I did!). I would say that EJB is by far and away the most covered subject and you need to know it in great detail. Despite spending an entire week dedicated to reading and making notes from the book I still didn�t know some of the answers (or I couldn�t remember them). This was because of the level of detail of the questions. I received the start of the recommended reading list from Sun. I already had Monson, Fowler & Gamma and to be honest this would be my recommended reading list and no more. You will need to know the Monson book back to front (this is particularly useful because it covers both 1.0 and 1.1 and details the differences). You will need to know the purpose of every pattern in Gamma as well as the consequences of each. As for UML (notice that there isn�t a UML subject score below despite it being a major element) you will need to be familiar with class and sequence diagrams at least. The breakdown at the end was as follows, this might help you to concentrate your studies (and mine for part II & 3) accordingly. Concepts83% Common Architectures50% Legacy Connectivity60% EJB86% (Better stop calling myself an EJB specialist now ;-) EJB container model100% (then again �) Protocols100% (Thank goodness for that subsequent question!) Applicability of J2EE100% Design patterns80% Messaging100% Good luck all. I�m off to hassle Sun for part II. Aaron Robinson Certified lunatic.
[This message has been edited by Paul Wheaton (edited June 20, 2000).]