Slashdot Mirror

As you said, the article doesn't really provide much detail. I am wondering though if this could be similar to what we are seeing with their recently acquired BEA WebLogic line. Have you read the BEA WebLogic 9.2 Virtual Edition offering? Most solutions will install products like Oracle and WebLogic on a dedicated server to maximize performance. What if they removed all but the necessary components of an operating system and optimized it? This has potential if it isn't, as you said, just a rebranded Proliant server with Oracle pre-installed.

Bea is doing some interesting things with their LiquidVM, which lets you run Weblogic as the "Guest OS" on top of your hypervisor. I can see this letting shared hosting companies offer J2EE to their clients.

the stack traces JRockit produces are formatted differently from those of Sun's JVM (which prevents my IDE from turning them into clickable hotlinks that take me to the lines of offending code). FSM knows why they made them different.

Since about two years back our stack traces look identical to SUN's. I'm at home right now and can't give you a specific release for that change. The difference was that old versions of JRockit used to show full method signatures rather than just method names in stack traces. We reverted to doing what SUN's doing because of people having the exact issues you're having. Which JRockit version (do java -version) are you using?

Oh and the line numbers in your stack trace will be wrong unless you turn off optimization - which is the the whole point of using JRockit.

If you have a repro for this, please post it here. Since you're running WLS you probably have a support contract for JRockit, in which case you should open a support case. Unless we know about the problem we cannot fix it :-(.

(It's supposed to be faster than Sun's JVM - I've never seen proof).

As always it's not provable for the general case as it depends on what you're doing. Two references that do have numbers for JRockit being generally faster than SUN's JVM are here and here, but those numbers may or may not be applicable to your app.

HTH, have fun :-).

the stack traces JRockit produces are formatted differently from those of Sun's JVM (which prevents my IDE from turning them into clickable hotlinks that take me to the lines of offending code). FSM knows why they made them different.

Since about two years back our stack traces look identical to SUN's. I'm at home right now and can't give you a specific release for that change. The difference was that old versions of JRockit used to show full method signatures rather than just method names in stack traces. We reverted to doing what SUN's doing because of people having the exact issues you're having. Which JRockit version (do java -version) are you using?

Oh and the line numbers in your stack trace will be wrong unless you turn off optimization - which is the the whole point of using JRockit.

If you have a repro for this, please post it here. Since you're running WLS you probably have a support contract for JRockit, in which case you should open a support case. Unless we know about the problem we cannot fix it :-(.

(It's supposed to be faster than Sun's JVM - I've never seen proof).

As always it's not provable for the general case as it depends on what you're doing. Two references that do have numbers for JRockit being generally faster than SUN's JVM are here and here, but those numbers may or may not be applicable to your app.

HTH, have fun :-).

At the risk of tooting my own horn, I blogged about similar material about a week before the Dark Reading publication. My blog focused more on the PR foul-ups that companies tend to make and ways to prevent those foul-ups rather than the technical response. It was based off of a recent Google vulnerability that got publicly posted as ?revenge? by the vulnerability discover who was unhappy about having not gotten enough credit for previously reported Google vulnerabilities. Neil Smithline BEA WebLogic Security Architect

yeah, i know it sounds like an advertisement, but it's my own personal opinion having used both products, and actively using it on protrade.com. I would personally discourage anyone from using hibernate and encourage a look at kodo for any large commercial/production site.

Yup, the core "kernel" is open and the ejb3 interface is open, but not the JDO interfaces, but they are both implemented by the shared "kernel".. sorry http://dev2dev.bea.com/pub/a/2006/02/interview-kod o-opensource.html

Bill Roth's blog at BEA has posted favorably about using BEA Workshop Studio on Ubuntu:

http://dev2dev.bea.com/blog/wgroth2/archive/2006/0 7/ubuntu_and_bea.html

He includes the following disclaimer:

Note: Please be aware that this is not an acknowledgement support, just merely my personal efforts with this software. If you think we should support Ubuntu, let me know.

Since when do standards bodies make for a higher quality specification? They exist as a political process to ensure a wider 'supported' specification & to ensure stakeholders get their say at the table.

The Java langauge spec is very well defined, and there's more than one reference implementation of a JVM available to study... which is why you tend to see a lot of separate codebase VMs for Java (Sun v. IBM v. BEA JRockit v. Aonix v. kaffe v. GCJ) and value added derivatives of Sun's implementation (HP has variants for Tru64, UX, OpenVMS, and NonStop, and some have OS-particular features). Compatibility wise, they broadly work well with minor issues -- see BEA's support matrix, for example.

Considering JRockit compiles all code into machine code before running it, that's a non-argument.

Everybody knows that going by car is faster than walking. That doesn't imply that some car is necessarily faster than some other.

Likewise, everybody knows that native code is faster than interpreted. That doesn't mean that gcc's native code is necessarily faster than JRockit's. Yet, that's the argument you're making above if I understand you correctly.

Since it's an independent implementation it contains different bugs from SUN's VM, and you may have better luck with it.

Or it could be that your VM process uses up too much memory and gets killed by the kernel's OOM handler. In that case JRockit will disappear even faster for you.

Disclaimer: I work for BEA.

Since it's an independent implementation it contains different bugs from SUN's VM, and you may have better luck with it.

Or it could be that your VM process uses up too much memory and gets killed by the kernel's OOM handler. In that case JRockit will disappear even faster for you.

Disclaimer: I work for BEA.

Generally I've found that you can tune the Sun Hotspot VM's generations to minimize pause times, though this requires a GC analyzer to determine what's taking so long. Ensuring objects stay in the eden generation, for example, ensures they get the faster parallel garbage collector -- the "stop the world" collector only runs on older generations. I've run heaps with a couple of gigs of RAM and it's pretty rare to see it pause for several minutes -- a few seconds, sure, but that depends on the application.

Another thought, have you tried BEA JRockit? It has both a concurrent generational and parallel garbage collector which you can swap at runtime, and also has a dynamic GC setting to optimize for "minimal pause times" (-Xgcprio:pausetime).

Unfortunately, it's only on x86, EM64T, AMD64, and Itanium for now (Windows + Linux), but a Solaris AMD64 and SPARC version is due in the coming weeks. It's free for general use, but requries a license if you want to use the management features / GUIs.

have to agree with Mr. Decaff there. Garbage collection has gotten more advanced using things such as incremental GC. Also, specialized VMs such as from bea (look for JRockit) use advanced techniques such as multiple GCs.

Also the article mentions:

For one, many of the now-large companies built from the ground up to operate on the Internet don't make Java a major piece of their tech strategy. Those include Google (GOOG) and Yahoo! (YHOO).

In this article, Li Moore (software engineer at Google) says: "Google makes extensive use of the Java platform. Large parts of popular Google products are written in Java. We also use Java in many internal systems and products under development." So, this would seem to be entirely untrue in Google. I wish people would do a bit of research, instead of speculate/lie (which are, of course, easier to do).

Finally, I notice that the author of the article asks Peter Yared and Marc Andreessen their opinion. What sort of answer would one expect from the CEO and chairman (respectively) of competing technology companies? Naturally, they would champion their own technology and bash a competitor. I certainly would if I were in their shoes!

+1. DWR is amazing. If you are forced to use Java, it is the way to go. At the risk of slashvertising, here is an article I wrote about using DWR: http://dev2dev.bea.com/pub/a/2005/08/ajax_introduc tion.html

Java programs use up so much at the moment that I doubt there's enough silicon in the world to produce them once Java goes 64 bit!!!

It is not the programs themselves as he JVM that takes up memory, and Sun has merged code from Apple that allows a single JVM to run all your Java programs. I'm guessing that will not stop the usual trolling though.

As for the other part of your statement, there already exists plenty of 64 bit JVMs, and as usual, my programs work fine without me having to change a single line of code:

http://java.sun.com/j2se/1.4.2/install-solaris-64. html
http://e-docs.bea.com/wljrockit/docs142/certif.htm l

There are more 64 bit JVMs available from IBM and others, for Windows, Linux, Unix...

It also depends on what you mean by "commercial". If you mean for commercial use, then yes. But if you mean that it is for sale... yes and no. You can use JBoss free (or could) and buy support. So really, the author is correct.

Proprietary is incorrect, though. It is not an applicable term to a J2EE app server. Or if it is correct, it applies to all of the other big app servers: Weblogic, Websphere, and Oracle. I can run these on many OSs and I can use different databases, different JVMs, etc, etc.

As to the review, anyone who has used one of the big name app servers for any length of time, would not find a single book to be a sufficient resource. It is good to see the reviewer point this out. Oddly enough, having had to use Oracles app server for a few years (and being dumbfounded at their horrid/incorrect/blatant lies documentation) I have found one of the best general resources to actually be BEA's dev2dev site. It contains good articles and detailed documentation on some of the steps others abstract away (especially concerning deployment descriptors for ejbs and jms).

Of course, Weblogic and Websphere still have the lion's share of the market... but I would love to see more JBoss (and less Oracle... ick).

And if your using Java on x86, don't forget to use jrockit from weblogic It's quite a bit faster than standard Java

And if your using Java on x86, don't forget to use jrockit from weblogic It's quite a bit faster than standard Java

No, the original point was that you said " guess the other thing is that Apache/PHP seems to scale better (or at least easier/cheaper [internetnews.com] than Tomcat." The article you link to doesn't mention java or tomcat at all. The presentation that is linked from the article doesn't mention tomcat but it does say good things about java/jsp/j2ee but they couldn't use it because freebsd's threading sucks. I should have known better than to start a discussion with someone that cites an article to back up his claim when it doesn't even mention what he's claiming.

"What would are some very large sites using jsp, which for all the links you posted, you haven't been able to show.More and larger sites use php than jsp. I pointed out ebay, which is still using dlls for much of it's heavy load. You brought up playboy, which seems to be mostly a static site running some cgi scripts: http://cyber.playboy.com/cgi/ab.cgi."

eBay used to use a dll. The dll is their old architecture, when they finish the migration the dll will disapear. eBay is a complicated site and the migration will take time. A lot of ebay is currently running on websphere. You can't easily tell because they don't use the .jsp extention for their files. That's a big site. They may have 1/3rd the traffic of yahoo but their profits and revenues are about the same. Playboy does use java. I think they use a mixture of JBoss and Tomcat. Just because a page says .html doesn't mean it's a static page. A lot of times, I've noticed this with JBoss especially, files look like html but they're not. You can see this at the sims online too, which also uses jboss. Look at this url from playboy http://www.playboy.com/magazine/playmate.html?sour ce=playmate_sectionfront since when can you pass query parameters to STATIC HTML? We used this technique with sites I've worked on as well where we set up our container to recognize .html files as jsps. If you look at netcraft.com for the server info you'll see playboy.com uses mod_jk whick is the connector to Tomcat. Sometimes it's hard to tell what sites use java since many times they'll use a servlet controller that calls the jsp's behind the scenes. Sometimes you'll see a .do which is the typical way to use Struts, but not always. Here's a bit of info on capital one they use java for a lot of their stuff. I think they use oracle's application server. B&H Photo uses java for their ecommerce site (Probably the most popular online photo/video retailer) but you can't easily tell from looking at the urls unless you know what to look for. bnh is the servlet context, controller is the main servlet handler and home is a page handler. Years ago they used to use Bluestones application server. Now it seems they're just using Netscape Enterprise Server. If your brouse certain sites with cookies disabled you can sometimes see a jsessionid in the url, that's a sign it's running a java web container. Verizon Online uses Java for their clients web email and a bunch of other support functions. Have a look at Bea's Client list among some other big names you'll find Amazon.com and FedEx. Ofoto.com, now Kodak Easy Share, uses jsp. Here's a story on jboss writing their own postnuke style cms in java because php didn't scale as well. These are just the few I could remember off the top of my head. And I've done my own benchmarks for m

Similarly, for ... business applications, enterprise servers ... you won't go wrong with Windows.

Why can't the community get togeather and create an open API like Microsofts Direct-X?

In any event, you don't usually play games on enterprise Linux distributions. So your post is rather out of place here.

I would say Linux is that OS. But it really needs support from the entertainment gaming industry to push is public support to the masses.

(And if your sarcasm detector needs some calibration, click on some of the links I gave before you post a comment based on my comment ...)

For good measure, let's look at the documentation from a J2EE vendor here.

While PEAK sounds intriguing, I'm not sure that major projects started by Fortune 100 globals will leverage a technology that lacks the level of documentation quality you can find in other products in that space.

I bring this up because documentation is often an indicator of the level of quality you can expect in terms of support. This is not to say PEAK is bad or poorly written, just that the supporting documentation and resources don't match those available for J2EE implementations.

Remember -- it isn't the best technology that wins, but the technology that is most accessible. In the case of enterprise APIs, even though PEAK may be easier and more scalable (and this is an excerpt from their page): But PEAK is different from J2EE: it's a single, free implementation of simpler API's based on an easier-to-use language that can nonetheless scale with better performance than J2EE. ...it will need some time and some nurturing in order to compete for mindshare with developers and non-technical decision makers.

The one thing that Java has going for it are "standard" APIs you can bank on. Is there a standard set of enterprise APIs for Python akin to J2EE?

And all of this isn't to say that one can't leverage both technologies where appropriate, even in commercial products...

a profiling Just in time JVM can profile the code while it is running and then next time around recompile the code to be faster.

That site doesn't list how the tests were carried out except to say that there were, no details on any profiling or optimization attempts that any serious deployment would use.

It does say that start up time is included too, so that taints the benchmarks.

oh, and he didn't even try jrockit

Here's a JVM that targets Itanium specifically.

Wrong market. WebSphere is an application server, not a web server. Yes, it has a web server included with it, but that's not its raison d'etre. You'd be better off comparing it to BEA's WebLogic Server or to the open source JBoss Application Server or the free Sun Java System Application Server.

In terms of market share, the last nubmers I saw had WebSphere first, with WebLogic close behind.

I'd also be very curious to see an independent mark of BEA's JRocket JVM, which is supposed to blow Sun and IBM away on X86.

I'd also be very curious to see an independent mark of BEA's JRocket JVM, which is supposed to blow Sun and IBM away on X86.

> Its' not the case in Java where it sounds like there's only 1 way to do each thing

I take it from this comment that you haven't actually tried java. You can "explorate" to your hearts content, and there are many ways of doing the same thing (some obviously better than others)

> mostly because of the plethora of APIs (WebLogic, etc.) that corporations force you to use over it...

Now this comment just has me bamboozled... You mean that Weblogic holds sway over you and force you to code in one way over another? Surely weblogic is just an appserver? Which runs code designed to the standard J2EE API spec? The same as using Tomcat , JBoss or Geronimo or even Hibernate? (All of which are free and opensource, and follow the same J2EE spec that Weblogic does -- they just solve separate parts of it, and can be combined to do it all if you require)... I fail to see how this is a corporation forcing you to use one method of coding?

Sure, if you are only going to look at one way of achieving your goals, then there is only one way to go...

Beehive is an open-source software project designed to deliver a cross-container, ease-of-use programming model and application framework for J2EE- and SOA-based applications. Beehive includes support for JSR 175 metadata annotations, the Java controls framework for creating and consuming J2EE components, a simplified Web services programming framework, and the Struts-based Java Page Flow technology for creating Web-based user interfaces and applications.

There are already some visual programming IDE's that are supposed to work at this gross chunked (fit them together like pipes) component level. I've noticed that they cluster around the B2B integration tool space.

Here are some examples. M$FT uses their acquired Visio engine in their biztalk server to model dataflow. BEA Systems does something very similar in their Weblogic Workshop IDE. Go to page 8 in this white paper of theirs.

Is this true programming? IMHO, its as much programming as writing a shell script.

"Microsoft was kind of pushing Passport for a problem that didn't exist..."

The problem of single sign on (SSO) does exist, particularly in the corporate world. Vendors implimenting Web Portals (MS SharePoint, Sun Java System Portal Server, BEA WebLogic Portal, Vignette Portal, etc...) have a particular interest in SSO and identity management via Identity Services to present a single interface to various systems in an enterprise.

My main problem with MS Passport is that it's Microsoft's version of a standard rather than a community standard. Applications can connect via MS's SDK rather than publishing the standard. Using Open LDAP, Sun's Identity Server, etc... will generally follow open standards and have better compatibiltiy to other open source/standard applications.

The benefit to Sun of GPL'ing their Java implementation would be expansion of their market influence. Right now, there aren't very many open source Java apps (comparitively speaking). This would change rapidly if a complete JVM/JDK could be included legally with every Linux/BSD distribution. Complete adoption of Java by the Open Source community would mean a sharp rise in the popularity of the language and this would help Sun tremendously.

Strange that BEA has it's competing (?) JRockit there too!?

Responding to the last comment in the parent post:

Actually, JBOSS has led the way in features, with Weblogic playing catch-up.

Versions of BEA WebLogic were fully released to the world before the first line of JBOSS code was even written - so how could JBOSS "lead the way"? (Aspects? *sarcasm*)

most people I know who have used both prefer JBOSS.

True in context, but useless information otherwise. Who are these people? Here is statement that is also true in context, but equally useless: Most people I know that have used both prefer WebLogic. Now here is something harder to argue with: BEA WebLogic's customer list at BEA's website. I invite all and sundry to post a link to comparable list for JBOSS.

Of course JBOSS may have its uses, being free, but it otherwise simply fails to directly match up with *any* commercial app server, not just WebLogic. And the failure is not just in features and customer base, but also in terms of testing, reliability, performance, long term support, J2EE certification, professionalism, and documentation. And these differences, for anyone that bothers to do even a small amount of research, aren't small.

Oh look, these people are patently under 14</sarcasm>

That in under a minutes googling. Now crawl back under a rock troll.

And for people who STILL believe that Java is slow, download eclipse (the url is above, for those of you can't google for it, or even guess it) and have a play...it's written in Java and *GASP* is not only a)amazingly useful, but b)damn snappy, seriously, those people who said Java would never work on the desktop because its too slow, they can start eating their hats.

Plus, I just did a job search for c++ vs. Java on a uk job site and guess what, the number of results returned was nearly identical, so by the look of things, lots of people and companies use Java. IMHO, for enterprise n-tiered applications its the shiznitz.

The reason that Java can be faster is because you can do some really nifty optimisations during runtime with the latest JIT compilers that are impossible (or at least monumentally difficult) to do with compile c/c++ code.

In closing: Here is another benchmark...maybe because he has a domain name you'll trust him more?

More Tuxedo information

My question is this: is there a place for proprietary (read 'closed') applications on said open/free platforms and frameworks?

Yes.

WebLogic

Acrobat Reader

StarTeam, Together Control Center

StarOffice

WebSphere

That's just off the top of my head.

Web services? OK...

Not sure why this should be a driver at a college, but there are of course plenty of Java web service implementations out there, including the Apache SOAP, UDDI and WSDL implementations that have reference status in the industry. If you want an advanced but friendly IDE for building Web Services in Java try WebLogic Workshop. I believe IBM WebSphere Studio and Borland JBuilder can also do wizard-like WS development.

As someone has mysteriously marked the parent as "interesting" it might be worthwhile to provoke a genuinely interesting debate below it so that readers are not too disappointed.

The parent insists that no Java IDE "can touch" VS for "any single thing you could possibly want to do", but a moment later admits that "VS.NET doesn't have as robust a feature set as some Java IDEs".

Features like refactoring, perhaps, as found on the free Eclipse IDE, or the modestly priced IDEA?

Or, looking a bit further afield, we could ask how one might develop a complete workflow system in VS, as you can in WebLogic Workshop?

My clients do these things all the time, but VS has a long way to go to offer a competitive alternative to the Java tools available now.

I program these types of systems for a living. Usually, everything I work on is HUGE. To put in in perspective, I recently finished the control system in the Chuang Residence. Alfred is very much into the latest and greatest - and had a 3 million dollar home theater put in (seats 7) as well as A/V distribuition (30 zones) throughout the house (61" HD plasma at the foot of his bed!).

So, it depends on the signal that you're routing.

Amplified: Great! it'll probably go as far as you need it. If you're going over a few hundered feet and don't care about the quality (think outdoor speakers), you'll want to look into doing with 70 volt transformers and matching speakers. (the higher voltage will drive the signal much farther, but the audio quality goes to hell). Generally with amplified signals the farther you go the quieter it gets.

Line level (non-amplified signal): Will often work just fine. Since it uses a much lower voltage it is much more succeptable to noise. The two biggest things to watch for are interference and ground loops. You get interference from power sources, running motors, wireless devices, etc. If you convert the signal to balanced it'll be less succeptable to interference (a balanced signal provides a ground reference). Small converters are available from several sources. Ground loops are caused when there is a voltage difference between the "ground" plane on either end of the wire. Often, you'll hear a slight 60 Hz hum. The PROPER way to eliminate this is to make sure all your equipment uses the same ground. If it's all on the same circuit in your house, then you're probably OK. If not, there are small inline devices to eliminate this as well. Look for "Hum Buckers" or "Electo-Optic isolators". Line level will go a suprising distance but you'll want a distribution amp to go over about a hundred feet. Keep in mind that the farther you go, the better chance of having noise problems.

Digital: Bleh! I really want this to work, but I haven't seen a successfull installation with more than 20 feet of wire! I worked one job where they tried both optical and copper but couldn't get it reliably (trying at about 75 feet of signal). When the signal degrades, the audio just goes away - not quiter or noisier. It suprises me that the optical signal won't go farther - it's light dammit! The issue seems to be that the TOSLink cables are designed around a plastic core that is easier to work with - but attenuates too much over distance. Having a device to rebuild the signal every 20 feet is problematic to say the least.

(sun won its suit against microsoft that it was unfairly squeezing out the java vm - then promptly sued microsoft for posting the microsoft jvm on windowsupdate.com because the license from sun didn't explicitly allow that. they won the suit and for some time windows users just couldn't get their hands on a vm. and if that doesn't decimate any gains from using java, i don't know what does)

What are you talking about? Windows users have always been able to get their hands on a JVM, just not the Microsoft one because of the lawsuit. Nothing stopping users from downloading one of the many other implementations of the JVM on Windows.

For web service development in the J2EE world, and soon for much else (portals, workflow management) you might like to take a look at BEA's Workshop.

I like its approach to messaging a lot - makes SOAP/HTTP just another transport like JMS, presents RPC and async alternatives very clearly and shows what's going on message-wise live, rather than being a separate code generator.

BEA is already pursuing this.

The last two versions of BEA's WebLogic Platform included WebLogic Workshop, which is a 'VB Like' tool to make developing web applications and web services much easier.

They even bought/hired a couple of ex-Microsoft guys to head the development effort.

Maybe that is why it looks so much like Visual Studio...

WebLogic Workshop

As has been said, IBM MQSeries / Websphere MQ is available for Linux. So is the granddaddy of all Message Oriented Middleware, BEA MessageQ, formerly known as DECMessageQ. So is BEA Tuxedo. So is Tibco Enterprise. I'm sure there are others but considering that the above four cover something like 99% of the MOM market, whether on Linux or not, it's pretty silly to say there are no MOM products available for Linux.

As has been said, IBM MQSeries / Websphere MQ is available for Linux. So is the granddaddy of all Message Oriented Middleware, BEA MessageQ, formerly known as DECMessageQ. So is BEA Tuxedo. So is Tibco Enterprise. I'm sure there are others but considering that the above four cover something like 99% of the MOM market, whether on Linux or not, it's pretty silly to say there are no MOM products available for Linux.

Sun didn't like Microsoft including their own version of the Java VM with their OS so Sun took MS to court to stop them doing so.

Microsoft removed the JVM from their OS, as per Sun's demands.

When a user goes to a web-site that uses Java, the user has to download a JVM.

Now Sun is complaining that MS doesn't include a VM with their OS? Sun wants MS to distribute Sun's JVM with the OS? What about other vendors that produce JVMs?

• Which IDE you want to use, ranging from Open Source $0 IDE to commercial ones.
• Which source control you want to use, ranging from Open Source $0 source control to commercial ones.
• Which middle tier implementation you want to use, ranging from Open Source $0 application servers to commercial ones
• Which database you want to use, ranging from Open Source $0 databases to commercial ones.

In short, you choose J2EE in order to have a choice of what software you want to use within your business and how much you are willing to pay for it, what hardware (Intel, Sun, IBM) and operating systems (Linux, Windows, Solaris, AIX, OS/400) requirements you have, and what requirements you have on the performance (single Intel box to 64 CPU Sun box to IBM mainframe) and scalability of your application.

You make J2EE match your requirements rather than force yourself to match .NET requirements.

No, eDocs has answered all my questions. :-)

I wish more companies would take the cost of putting all their docs online, like BEA and Oracle do...