Win a copy of Spring Boot in Practice this week in the Spring forum!

Sanly Fang

+ Follow
since Jan 31, 2005
Cows and Likes
Total received
In last 30 days
Total given
Total received
Received in last 30 days
Total given
Given in last 30 days
Forums and Threads
Scavenger Hunt
expand Ranch Hand Scavenger Hunt
expand Greenhorn Scavenger Hunt

Recent posts by Sanly Fang

I started preparing the exam by reading RMH book (java web services), and I do think this is a must read for the guys with little on-hands experience on web services technologies like me. I read the book very carefully even if the topic is outside the exam objectives, anyway, I was trying to learn web services systematically rather than just got sun-certified.

Besides this book, sun blueprint is a must read.

For higher score, you may need to read other resources, for example, WSI BP1.0a, j2ee tutorial (relevant chapters), guide, mocks, MZ's guide & quiz.

I have read all books and resources that mentioned above, some of which I read twice.
[ May 16, 2007: Message edited by: Sanly Fang ]
Dinesh, I don't think it is required to remember all xml details, but given a specific feature, you need to tell where it is configured, for example, which xml file is used for setting up handler chains? Also it's necessary to know the basic structure of the config xml, for example, element order, what's the parent, child. Anyway, it seems practical to remember details about WSDL but have a general knowledge on the structure of deployment descriptors, mapping files.

design pattern questions in the exam is very simple and straightforward, it covers only 6(?) patterns mentioned in the exam objective

Client side: business deligate, service locator, proxy
Server side: web service broker, facade, adapter

That means, what you need to do is to understand the question, determine which side(client or server?) the question is asking, then pick up an answer from the three.

If still not confident with the design pattern section, repeat mock exam again and again, that will make you confident.

Good luck!
hello Dinesh,

For JAXB, JAXR, JAXP, you need to be familar with the API, at least the basic steps for manipulating an XML document, also the high level concepts may be questioned in the exam.

The security questions will be asked at a very high level, for example, you may not be asked the implementation details of a specific security specification, but you should know the purpose and general functionalities for each security spec mentioned in the objective goals, for security section, I think reading MZ's guide and guide is enough.

The design questions may be in the form of a scenario, it seems the most relevant reading material is sun's blueprint.

No questions on mapping xml details in my test, but you need to know where to configure (in which xml file) a specific kind of information and the deployment structure (where to put the config files).
I passed the exam today at 98%.

I missed one question in "endpoint design and architecture" section.

Though the score is nice to me, I think the test is more difficult than what I had expected. (for me, the preparation time is longer than the time I spent on SCEA part I)

This is because when I am working on mock exams, for example,, it is likely that I can point out the correct answer(s) immediately, but in the real exam, for several questions I should study each choice carefully, and guess a best answer.

4 or 5 drag and drop questions in the exam and many questions on security.

The exam do cover some level of code details and APIs, but appear only in choices and drag and drop items.

I don't think it's required to practise writing code when preparing because it may make me distracted to vendor specific details (for example, axis), I must confess that I haven't written a single line of code using web service technologies till now.

Repeat mock exam again and again because this will make you confident that you'll get a score higher than 90%

Read MZ's guide & Quiz thoroughly, which make me confident that I should have a 95+ score.

I can't open this link for some days, some problems from the site or wrong url?
1) System responds a list of available flights(Products) as search results, customer makes choices and a new itinerary is created in memory(Order), which is then be paid and saved...

2) CMR/CMP is good when saving data into database, while DAO approach is better to be used when getting tabular data from database for display purpose (for example, searching), related pattern is ValueListHandler.

Originally posted by kundan varma:
Hi Sanly
I still have few questions, hope i am not irritating you.
1) what does itinerary means? I thought earlier that it is all the data which comes in response to proposed itenerary. But some body told me that it is the selected data out of those available schedules which is referred to as itenerary.

2) If i use local interfaces for EJB's that is no marshalling-demarshalling business, then which approach is faster or better while accessing display data. Is it directly accessing EJB's using cmr OR using DAO pattern ?

Your views on that....


It's good to let app-client access EJBs directly, many other guys (including me) opt for this solution.

Now the question becomes "how to justify your choice".

If you search this forum, you will see several answers, for example, make an assumption that app-client and EJBs are in the same intranet...

Actually, I don't think this is a serious issue because it only affect your assumptions.

For me, I make an assumption that app-client access EJBs via RMI-IIOP over SSL, and if firewall exists, the firewall permits RMI-IIOP over SSL protocol or else tunneling mechanism is applied so that the application client can call EJBs directly.

In summary, it's safe to keep things simple (simplistic, anyway, this is an exam) for this kind of issues which are not clearly described in the requirement, just put some words in your assumption section and that will be ok.

Originally posted by kundan varma:
Hi Sanly
what approach you followed for connecting application client. I am thinking of connecting directly to EJB for performance reason. But what if rmi port is blocked ? they may block it for security reason. What can be the reason for a company for not blocking rmi specific port? I will search in the forum but at the same time it would be very helpful if you can answer it, or can throw your view on this.


If you choose the approach that client accesses ejb directly, you need to maintain session state in business tier or else code will be duplicated between client tier and presentation tier.

Regarding security issue, app-client could access ejb via RMI-IIOP over SSL.

You can get further information on "session management" and "RMI-IIOP over SSL" if you search this forum.

Originally posted by kundan varma:
Hi Sanly
I am thinking of application(swing) client to directly connect to EJB tier.But i am confused how to maintain session and security at that time.what approach you followed. Can you give some suggestion ?


1. 20-25 classes.
2. I make an assumption that a segment is a reservation for a specific flight. Itinerary/Segment/Flight is analogous to Order/LineItem/Product.
no "ticket" in my class diagram.

Originally posted by kundan varma:
HI Sanly
How many classes you had in your class diagram? Do you think there is any relation between reservation and ticket?


The patterns are in my component diagram (and supporting document).

Basically, the name of the component contains two kind of information:
1. What is it?
2. What's the pattern behind the scene
For example, ***DAO, ServiceLocator,...

On the other hand, the stereotypes hold information on how the component is implemented in term of J2EE technology, for example, <<Servlet>>, <<JSP>>, <<LocalEntityBean>>,... etc.

Originally posted by kundan varma:
HI Sanly
How you represent design pattern used in your assignment.Is it in component diagram or sequence diagram ? There you show <<pattern name>>
or <<class name>> then in note which pattern.


I am always thinking of a simple solution because I am afraid complex/comprehensive design will add burden to the guy who assess my assignment, in addition, given so many assumptions and details, it's easy to let him/her misunderstand my ideas.

Most importantly, why we must complicate things if simple solution can gain high score? Anyway, it's an exam, why an architect should make so many assumptions on business requirements in reality?

Originally posted by Terry McKee:
I agree that it is important to keep things simple, but I think it would be a mistake not to account for this scenario. Even though it is not explicitly in the requirements, I think it is safe to make some well-thought out assumptions concerning this (just document them). In the real world, a customer could simply be purchasing trips for other people. This happens all the time. I think that I am going to have something that encapsulates the person that makes the purchase and a separate something that encapsulates the person or pepole that actually fly.

One more thought, think about how frequent flyer programs work in the real world. Does each passenger get frequent flyer miles or does each customer? I believe it is the customer.

You don't have to, actually it's your choice to decide which objects in your design are important, and it's the important things that will be on your diagram or otherwise the diagram will become full of details and lead to messy style.

For example, in my component diagram, no specific JSP component for each page, instead, there is only a single component which stereotyped as <<JSP>>.

Originally posted by kundan varma:
HI Sanly
Do i have to relate these classes in the component diagram also.I mean mention those class in the class diagram in component diagram. What i mean is suppose i have a class Reservation, then do i have to show reservationEJB, ReservaionJSP in component diagram.


I may not be able to give you help on class diagram since I lost all my 5 points in this part.

In general, the diagram is good as long as it meets all requirements, two aspects are important: UML semantics and OOAD.

Originally posted by kundan varma:
THanks sanly
one more question. If i make class diagram tech independent then it will be very similar to BDM classes already provided with 2-3 additions.
I am still a bit confused. I have gone through a lot of docs but still not sure. Any hint to proceed in some direction without violating sun's law is most welcome.


You'd better make your component diagram pattern-oriented, which means, keep component in such a granularity so that everyone else can tell easily which patterns you are using and how they are structured.

Regarding class diagram, technology independent is a safe choice, as many others including me has done in this way.

Originally posted by kundan varma:
HI All
I have one question will FilghtSessionBean type of class goes in component diagram or class diagram. I am a bit confused which matter goes inside class diagram and what under component. Does component diagram just shoes JSP,SESSIONBEAN,ENTITYBEAN OR real class.

Thanks for suggestions

I cannot remember which resources I am using for additional reading because there is a long time after my Part I. There are some useful links in yahoo groups describing which resources are must-read, and of course, you'd better search this forum to see what other guys said.

Yes, I mean Cade's book.

Originally posted by Mark Cave:

Do you mean Cade's book?