In this interview, Thomas Kurian, Senior Vice President of the Oracle 9i Application Server, discusses the future of the J2EE platform, and provides recommendations on improved Web services support. He compares application servers with TP monitors, talks about integration servers, comments on the validity of modern marketshare surveys and compares Oracle's product offerings to those of BEA's and IBM's.

Watch Thomas Kurian's Interview

In his interview, Kurian states, "against BEA, clearly on the J2EE side, we have much better performance and scalability than WebLogic." Nothing seems to be further from the truth.

Kurian is probably referring to their recent benchmark in which they achieved the highest performance numbers on SPECjappserver2001. Check out http://biz.yahoo.com/prnews/030602/sfm084_1.html.

First off, why did they use an outdated benchmark, which is based on J2EE 1.2 to achieve their results instead of using SPECjappserver2002?

In any case, the result is actually quite poor if you look at it. Oracle simply threw a lot of expensive hardware at the problem to achieve better throughput, which any vendor could do. If I'm looking at the results correctly (see BEA's results on SPECjappserver2002 at http://www.spec.org/jAppServer2002/results/res2003q2/), Oracle used 50% more application processors and 50% more database processors and only beat BEA's score by 13%.

The bottom line is that BEA still has the leading performance results according to the latest benchmark numbers in the multinode category.

Even if you look at Oracle's database benchmarks, they use Tuxedo on the front end -- see http://www.tpc.org/tpcc/results/tpcc_perf_results.asp?resulttype=cluster&version=5.

Frankly, I'm tired of Oracle's lies.

Couldn't agree with you more! Oracle's much touted app server is going nowhere, so they have to resort to making tall, unsubstantiated claims. I am continually amazed at their gall in making such claims. As anybody knows, with a few clicks of the mouse the truth can be easily seen. Let's face it, their core market (db) is being trumped by IBM, their appserver isn't really ready for prime time, their apps are losing market share ...

But ... Oracle provides good entertainment value. Witness the recent bid to acquire PeopleSoft :-)

A BEA bid is next.

Now ORCL wants to buy PSFT, the chance of it buying BEAS is greatly reduced. It is very difficult to handle 2 large acquisitions at the same time. I think HPQ will be the beneficiary.

Der Rockfart, I hear Wall St is looking for a few jokers like you to sell bridges.

Unfortunately their marketing is quite effective on middle management who really don't understand what's going on. From said group I get (quite literally) "BEA - never heard of it" - something I don't find entertaining at all :). My life would be easier if Oracle, Sun and BEA merged. Either that, or BEA needs a major marketing push to gain brand name recognition.

cheers,
Markus

No one would dispute that Orion Server is a faster both EJB & Servlet container than Weblogic. Why do you think a different sticker on the box changes that? I can't answer scalability, but I do know that using java to scale isn't the answer if performance is a concern.

I want to know why on earth a lot of profession take the performance like the unique importante face of an application.
There are a lot of faces like scalability, maintainability, easy of use, performance, support, etc,etc,etc. I can't negate that performance is good, but performance without all the other faces of a professional system is nothing.
Probably (Really i dont know) oracle is faster than the competition, but it is really that it has a lot of bugs, like the points write before.

Thanks
German

First make it work, then make it fast! ;)

Having experience with BEA WLS, JBoss and ORCL one can say OC4J even in its latest releases is really buggy and unfriednly for developer. Have never been so angry at some software product than Oracle's wannabe J2EE containers.

If you want to save your nerves then - DO NOT USE IT.

Interesting, because I've used Orion Server and OC4J (derived from Orion) for several years and found them both to be quite good.

I just finished one project on OC4J 9.0.2.. not the latest, but its not always good to change versions in the middle of development.. and the app server gave us no problems at all. My only complaint would be the irritating process you must go through to build custom JAAS modules, but then, its exactly what was taken from Orion. It looks like if I was using a full 9iAS, it may have been easier.

I too have used BEA, JBoss, etc.. and can't see what the problem is here.. Is it that people are so used to the quirks of one server that they decide all others are just not as good? All app servers have good and bad points, but if you are used to them, its easy to gloss over. There are thinks I like and don't like about each, but I'm not going to say don't use any of them because of that.

I would like to hear of *REAL* problems.. not just complaints. Did the server crash on you? Did it corrupt data? Did it not adhere to the spec APIs?

Well - self-referencing CMR relations returning source object instead of the referencing one, not supporting Timestamp objects in ejb-ql, not being able to use composite primary keys, client side transaction management - only to name a few.

Previous versions were even more frustrating ...

>>Previous versions were even more frustrating ...

Most of these issues are addressed in OC4J 9.0.4. There is a Developer's preview available. I will encourage you to download and test this in OC4J 9.0.4

http://otn.oracle.com/tech/java/oc4j/904/content_preview.html

regards
Debu Panda
Oracle

are you really sure about what you are saying!!??
i'd just invite you to test better the product........really better....i use iAS/oc4j
almost everyday with NO problems .... iAS/oc4j becomes better day by day........
so i think what you are saying is TOTALLY untrue.
thanks
Paolo

oracle does not have a clear path of who will administer their J2EE server products. They have been tying the things so much to database thinking the existing DBAs can become J2EE/Web server admins overnight. This has happened in 11i upgrades also. I have seen existing DBAs installing webserver,apache,SSL etc without having any clue of what are they.

Second thing is to take a clear cut decision on products like Forms, Reports and move/eliminate them towards J2EE, since bulk of the oracle's development platforms are on them. Moving them to 100% J2EE will give them a solid userbase of J2EE applications.

Keep the app server light and not a 3 GB install product, with non-J2EE products like discoverer, reports, forms etc.

Sundar