On jboss.org, JBoss founder and lead developer, Marc Fleury, responds in bulk to questions he receives every day concerning JBoss, JBoss Group and his vision of the future of the J2EE marketplace. According to him: "With JBoss 3, we are going orbital. JBoss 3 makes the JMX view stronger, simpler and more central."
"EJB almost takes a back seat to this fundamental SYSTEM view of the world that we are pushing. We push JMX to provide a real 24x7 microkernel architecture, a container of containers, the super-container or the super-server, as we like to call it. "
Read more at http://www.jboss.org/vision.jsp
Marc, you're a true son of a GUN !
Any company (M$haft) included who ignores the OSS movement is destined to DIE a painful death
Expect to see a 'Hard Core Tech Talk' Video Interview with Marc Fleury here on TSS real soon! :) Rumour has it, it's going to be posted sometime this December.
Now that is what I call true killer instinct.
"Basically, I think when Rabbit Hole comes out, our technology is going to blow everybody else out of the water. My goal is for JBoss to become the third OpenSource project to massively crossover into the corporate IT sphere after Linux and Apache"
One of my wildest dreams is to be one among the esteemed team of JBoss developers.
I can see some good things about JBoss. First and foremost, the dedication of the developers, and then the neat architecture. However, I sincerely doubt that BEA will be packing it in next year, as Marc claims.
The reason lies in the psychology of IT managers.
The reason is that managers do not care about technical sophistication. What they do care about is features that allow them to keep their jobs, and those features are 1) reliability, 2) scalability, 3) on-site support, 4) J2EE compliance. That means that to be used by big corporations JBoss should stop refactoring its architecture and start concentrating on load-balancing, distributed caching, VC funding for a large on-site support force, and convincing Sun to certify JBoss as J2EE.
If you ever noticed, Linux and Apache have exactly the right kind of priorities. Torvalds' top priorities are probably stability, scalability and portability. That's why Linux is not a microkernel. It was built to be used in production from the very beginning. Apache is just now starting to add threads. Why? Because the Apache group just wanted a web server that worked, so they chose to use processes, which are a little inefficient but are portable and stable. And finally, now that everyone uses it they are adding threads.
JBoss is a lot more like the GNU Hurd and PostgreSQL - a project that is always interesting and always innovating but that corporate developers are afraid to use.
It will be interesting to see what happens. EJB Containers are going to be (if they aren't already) become commodities.
They are going to become like TCP/IP... so its the stuff that will sit on top of that that will matter even more.
Both Marc and Rickard make some ballsy comments don't they... "We are going to take over the world!". We will see.
I want to see more tools that I can use. I don't expect to ever look at an XML deployment descriptor. Tools like EJBGen, various IDEs are OK... but they can get better.
For the record, Marc is the one doing ballsy comments, not me ;-)
I don't say "we'll take over the world" as it is not a any personal priority for me, neither is it really possible nor desirable.
That said, I hope to be able to continue to provide y'all with good tools that makes your development easier and more efficient. But to say that they will take over the world, well that's not for me to say :-) (ok ok, so sometimes I do that too, but only when I get overly enthusiastic about my projects... btw, have I mentioned that AOP is cool? No? uhm.. it is.. you'll see..)
Rickard/ "I do not want to take over the world"
Aw come on, Marc just has a bigger mouth, that's all.
You guys are all pretentious egomaniacs. Why else would you be working harder than anyone else for zero pay in Open Source?
Marc is an annoying publicity whore, who's intolerance of anyone he does not perceive as a deep technologist is immature and does a disservice to JBoss. However, at least he's honest about his ambition.
Personally, he comes off as a total jerk. He should just pump out top quality code and keep his mouth shut.
"You guys are all pretentious egomaniacs"
ROFL :-) Man, that one was funny ;-) Yes, I understand that some people will never ever understand why someone would work harder than anyone else for zero pay. Kind of a mystery, isn't it? ;-)
My personal ambition is to make the best software there is for the categories of software I decide to work with. And it's not the end result that is the really interesting part, but getting there. That's what most people who don't understand OpenSource just don't get; building the damn thing is the wildest ride you'll ever have. And it's probably impossible to understand this unless you've done it yourself (successfully).
/Rickard, back to coding for zero pay.
>...it's not the end result that is the really interesting
>part, but getting there. That's what most people who don't
>understand OpenSource just don't get; building the damn
>thing is the wildest ride you'll ever have.
I just want to say that I believe this statement settles my unanswered question about priorities.
P.S. Congrats on the new job ..
About reliability and scalability. Refactoring architectures is probably one of the best ways to achieve your mentioned goals. It is a known fact that more code=more bugs. Refactoring generally tends to lessen the amount of code needed to solve a particular goal, hence "automatically" making it more reliable, in terms of code bugs. Also, refactoring architectures to make them simpler is also a great way to enable generally applicable functions, such as scalability. It was interesting to listen to Larry Ellison this past J1, as he bragged about the new Oracle9 implementation being incredibly fast and scalable since it was so small and lean and mean. Knowing that he really was talking about Orion, I also knew that it to a large degree meant having a small codebase based on a good architecture.
JBoss 3 is about the same thing: doing more things with a smaller codebase, and more simplified and generalized architecture. This *will* lead to better reliability and scalability for the abovementioned reasons.
>JBoss 3 is about the same thing: doing more things with a >smaller codebase, and more simplified and generalized >architecture. This *will* lead to better reliability and >scalability for the abovementioned reasons.
But this begs the question of priorities. What do you do first? This matters because it determines market share, and market share is also an important issue for an IT manager.
Guglielmo wrote :
However, I sincerely doubt that BEA will be packing it in next year, as Marc claims.
The reason lies in the psychology of IT managers.
I completely agree with that sentence. IT Managers are generally sitting between two chairs : IT (architects, developers, etc.) and marketing.
While IT pushes for a technically good solution, easy to use, standard, open, etc., the marketing pushes for solutions that will make the company look good and earn money.
If a marketing guy has a choice between a technically good JBoss and a not so good WebSphere or BEA, they will choose Websphere or BEA. Nobody ever got fired for choosing IBM. And God knows IBM has made bad products...
That will allow the marketing to make announcements like "We support BEA" or "We're partners with IBM". The clients will think "These guys are serious.". Moreover, BEA or IBM will also promise them full support, they will tell them that IBM will promote their solution, etc.
Of course, this will probably not happen, and the IT team will have to struggle with a product they don't find very good, until the project is finally cancelled because it's too buggy or slow...
Well, you get the idea. Even bad commercial products are hard to replace by good open-source products. And, after all, BEA is not so bad ;-)
On the theme of "nobody ever got fired for recommending IBM"
Basically, if you are going to be realistic about the J2EE marketplace, there are only three relevent app servers out there: Weblogic, WebSphere and JBoss. I just don't see the user and developer base on the others to make them anything more than pet projects.
As a consultant who's worked with all three, this is my personal experience.
JBoss: scalability and stability are the best out there. Depending on the app, I've seen JBoss run between 300 and 1000 throughput invocations per second. Anything below these numbers was because I shot myself in the foot with bad packaging, misconfiguration of Tomcat, etc. Granted I had to figure these out by myself. The documentation and free support were lacking. I hope to see improvement in the for-pay documentation model that JBoss is pushing these days. I also hope to see the network of consultants around JBoss offer the kind of support that everybody needs in the Enterprise World.
WebLogic: Not a bad product. Stability problems though. It's kind of bitch to configure and taxing to keep up. In the last installation I worked on, it crashed about once a day.
WebSphere: Between the two proprietary leaders, here's a clear case where this product needs serious re-factoring. Not that we can see it, but I think the code base is a slowly decaying pile of atomic waste. I have a box running at a client install, just jsps, I don't want to add anything to it for fear of explosion. Man, it is really clunky software.
Back to "you never got fired for buying IBM...," I replaced the previous consultants at my last client who did get fired for recommending WebSphere. It was impossible to use in development. You have to understand that the end client doesn't care as much about brand as you think. They care about the solution being cheap and stable. If you find a solution like JBoss that wins on both fronts, you're bound to see it win big accounts. The people who love JBoss are the tool vendors and consultants because we get to keep the extra money rather than paying it out to the middleware vendors.
Be wary of "what the IT manager wants." He doesn't know what he wants.
I think Sun, HP, Oracle, SilverStream, Borland and Apple would beg to differ with your argument for only three relevant app. servers. Companies like Oracle and Borland certainly don't view their app. server strategy as a "pet" project.
IBM has certainly had many considerable victories with its eBusiness strategy. WebSphere happens to be one of many tactical pieces in this strategy. While WebSphere has some idiosynchracies, you will not easily pry it out of the hands of many of the top companies in the world.
JBoss is a joy to install and simple to begin working with, but, it would be naive to think that they will replace IBM, BEA, Oracle and Microsoft from major IT installations without a world-class fight.
<i>I think Sun, HP, Oracle, SilverStream, Borland and Apple would beg to differ with your argument for only three relevant app. servers. Companies like Oracle and Borland certainly don't view their app. server strategy as a "pet" project.</i>
They and their 200 total user base probably love their product. Doesn't change the fact that the market is consolidating down to three major players.
<i>JBoss is a joy to install and simple to begin working with, but, it would be naive to think that they will replace IBM, BEA, Oracle and Microsoft from major IT installations without a world-class fight.</i>
Well said. It's not gonzo attitude alone that will establish JBoss as the "open monopoly" Marc likes to talk about. That will only happen when you have a strong independent network of consultants who push and support the product.
Posted by Chip Tyler 2001-11-29 10:40:24.0.
> JBoss: scalability and stability are the best out there.
I find it very difficult to believe. Would you provide facts to support this claim?
> Depending on the app, I've seen JBoss run between 300 and
> 1000 throughput invocations per second.
I took a look at production site references that are posted at jboss.org. None of those references appear to have links to actual live sites. Some sites that I was able to find appear to be very low-traffic sites. If you can, please provide a link to a site that would support your claim of 300-1000 invocations per second or anything high-volume at all.
I took a look at production site references that are posted
>at jboss.org. None of those references appear to have links
>to actual live sites. Some sites that I was able to find
>appear to be very low-traffic sites. If you can, please
>provide a link to a site that would support your claim of
>300-1000 invocations per second or anything high-volume at all.
I have a theory that many people's experience with JBoss is as follows:
1. They find out that there is a GPL-licensed EJB server called JBoss.
2. They read Marc's comments about how great it is. Like the comments on the mythical scalability or such statements as "the game is over", or "Bea's days are numbered".
3. All excited, they download the product, and look for some evidence of the claims made on the web site. All over they find references to JBoss' great architecture, the fact that the best developers in the world work on JBoss, the superior installation etc..., but they start gettting nervous because there are no recommended hardware configurations, no benchmarks, no mention of a coherent cache, no mention of multiplexed tcp sockets, and so on.
4. They figure, it's just a matter of time, that sooner or later the really important features will be there (like Rickard was saying above), and they wait.
This reflects my own experience. The final result is thousands of developers with JBoss sitting in a drawer. Put another way, Marc's savvy marketing strategy is artificially inflating the download numbers.
Now, ask yourself: if hundreds of thousands of people evaluate a product but then don't use it, what does that tell you?
Can't speak for others, but we tried it, and now we're using it. That is on a number of projects for different clients and research projects. That doesn't mean we use for every one of our J2EE projects, but some of them, and they are large scale projects. Most specifically, we plan to use JBoss for the Oregon Coalition for Interdisciplinary Databases (OCID) and other projects at the Northwest Alliance for Computational Science and Engineering. Our tests have shown JBoss to be just fine a a number of architectures. That doesn't mean others won't work well too, we just found JBoss a cut above, for our purposes (that being the key phrase in that sentence). I know everyone here has their preference, but our jobs (usually) is to choose the best tool for the job, not advance a personal agenda.
As for Chip Tyler, I'm not sure where the anger comes from, but I'll submit some comments. I've never worked for an open-source project, but I have worked for start-up companies, and I know what it's like to work long hard hours for crumby pay. You don't sleep at night when you worry about paying people, and whether or not you'll still be in business next month. But when it succeeds, it feels great. YOU have just BUILT something of your own with a group of like minded people. You feel like the 80's USA hockey team when the beat the powerhouse Russians. Man, it feels good. It's not about ego, although there is always some involved. It's about the process, about seeing you're baby grow from inception. I don't know where the bitterness comes from, but I don't think it's welcome. This is an intellectual forum, not a thread system on the politics of success. If you are not of like mind, just say so, but I don't think there is call for being insulting.
Can't speak for others, but we tried it, and now we're
If you can spare the time, could you say something about your architecture and the workload and load you plan to have?
>As for Chip Tyler, I'm not sure where the anger comes
>from, but I'll submit some comments.
I think Chip's feelings are pretty widespread. I think it's irritating when open source developers explicitly claim to be world experts, and expect you to treat them as such. The natural reaction is "who do you think you are?".
For example, the JBoss web site used to claim that Rickard is one of the world's greatest experts on EJB, but that really depends on your point of view. After all, Rickard himself wrote that he worked on JBoss because he wanted to learn how implement an EJB server. That means it was a learning process. Now, it seems to me that one cannot be learning and be an expert in the subject at the same time.
There is also something to be said about modesty :)
But it still escapes me why arrogance provokes anger (as it undeniably does), so if someone can point out the connection, I would be interested to read it.
"For example, the JBoss web site used to claim that Rickard is one of the world's greatest experts on EJB, but that really depends on your point of view. After all, Rickard himself wrote that he worked on JBoss because he wanted to learn how implement an EJB server. That means it was a learning process. Now, it seems to me that one cannot be learning and be an expert in the subject at the same time."
Disclaimer, I am a definite JBoss fan. With that in mind, I think most people would agree that Rickard, Marc Fleury, Scott Stark and other top JBoss developers are some of the world's leading experts on J2EE. In fact, I've heard that BEA tried (unsuccessfully) to hire Rickard several times.
All work, especially in a rapidly evolving technology field is about learning. The only way to stay ahead of the curve is to constantly force yourself to learn new things all the time. If you are not learning, you probably aren't trailblazing either. Thus I see no contradiction between having chosen a project to "learn" and becoming an expert in the process.
"But it still escapes me why arrogance provokes anger (as it undeniably does), so if someone can point out the connection, I would be interested to read it."
I think the anger is probably provoked by professional jealousy on the part of others. Success always attracts jealousy. See Microsoft hatred. Having attended one of the JBoss trainings, I was very impressed with Scott and Marc. It was easily the most technical training I've ever attended. When it comes down to the technology these guys deliver. I think the tone people are commenting on is just showmanship. In fact, if you follow Open Source at all (compared to Slashdot, the Serverside is pretty sedate) the tone seems to go with the territory. In fact, Rickard apparently was amused and I doubt Marc would take it personally either.
Having attended one of the
>JBoss trainings, I was very impressed with Scott and Marc.
>It was easily the most technical training I've ever
>attended. When it comes down to the technology these guys
What was the training for? To learn how to write EJBs or to learn the internals of JBoss?
Our trainings around the world are mostly on EJB on JBoss (2 days) with the other 3 days being JBoss internals.
These trainings are targeted at advanced developers and focus on JBoss the architecture, the code, the vision. Heady stuff, really succesful so far.
I will plug our site here
. Our next training is in January in Boston, Feb in Sydney and March in London. You also need to take the training to be certified as a JBoss affiliate contractor.
I usually don't participate much in threads, but here goes. I did find the "pretentious egomaniacs" funny and right on the money, you should see what they said on slashdot I got **ripped open**, hey that is part of the game. When I was 15 I got grounded because, while I was #1 in my class, my history teacher put "pretentious and un-pleasant" (eleve pretentieux et deplaisant in french) on my report card and my parents got so pissed off they grounded me in my room. But see I couldn't stay my room, I was coding this *awesome* game on the ZX81's awesome chip the Z80, called "death mission", we were going to make an adventure game fit on 16k of RAM all coded in assembly, rightfully it never saw the light of day. So I would sneak out at night and go coding under-ground with decoy pillows in my bed.
To make a long story short, I eventually got caught by the police for being on the streets at 5 am and under-age, my parents couldnt really believe i wasn't going out to clubs and smoking dope (that came later).
coding for zero pay and *still* pissing off authority...
But see I couldn't stay my room, I was coding this
>*awesome* game on the ZX81's awesome chip the Z80, called
>"death mission", we were going to make an adventure game fit
>on 16k of RAM all coded in assembly,
Umm. When I was ten I wrote some assembly code on the Z80 as well. I had a Sinclair Spectrum, the one with the rubber keys and 3 colored shift keys, so you could type code at immense speed (unmatched I think by any IDE developed since.) The Zilog 80 did have a good assembly, but ironically I thought the 6510 (which if memory serves me right was custom-made by Motorola for the C64) was cooler. We used to have programming contests with a friend of mine who had the Commodore.
Life was much simpler then.
P.S. I guess this thread is pretty dead by now, so you'll forgive for the lighthearted message ...
OpenSource developers often a different "currency". Some prefer fame a.k.a. worship (apparently some examples leap to mind here but please don't forget R.Stallman himself who seems to have copyrighted -- oops lefted -- the concept). Others prefer the joy of working in their dream job (just without the pay). Some like to see their name in a product, and others like the "community" and the "religeous" aspects.
Whatever. If JBoss can eventually take its place with FreeBSD and Linux (oops I meant GNU/Linux) and Apache, well good for it. More choices don't hurt.
Disclaimer, I am a definite JBoss fan. With that in mind, I think most people would agree that Rickard, Marc Fleury, Scott Stark and other top JBoss developers are some of the world's leading experts on J2EE.
There is a saying that goes "99% of the best developers in the world do not work for you". JBoss is no exception. There are a lot of very smart and expert J2EE people out there. Some work for major J2EE vendors, others for minor J2EE vendors and some are not even in the J2EE market at all. You just don't hear about them.
Ignore this simple fact at your own risk (or your company's).
". Put another way, Marc's savvy marketing strategy is artificially inflating the download numbers. "
Totally agreed. The reason Giga et. al measure license revenue is because that is the best reflection of what's actually being run in production. For example, I've probably downloaded JBoss about a dozen different times - to try out tutorials, or to work with other products, etc - but I'm not running it in production.
My guess is that if you looked at production installations (and by production I mean a real company - not someone's personal site) - that JBoss has less than 1% of the market share. Hardly game over.
Basically, if you are going to be realistic about the J2EE marketplace, there are only three relevent app servers out there: Weblogic, WebSphere and JBoss. I just don't see the user and developer base on the others to make them anything more than pet projects.
Being a supporter of Borland (see my review in the Reviews tab), I have to disagree with this one ;-). I think a whole lot of projects are done with other app servers, and not just small ones (Intranet apps, etc.). Just seeing that Borland, for example, has been chosen by Cisco, major banks, etc. makes it a valuable product, IMHO.
WebSphere: Between the two proprietary leaders, here's a clear case where this product needs serious re-factoring
Back to "you never got fired for buying IBM...," I replaced the previous consultants at my last client who did get fired for recommending WebSphere
You have to understand that the end client doesn't care as much about brand as you think
I don't think the end client cares so much. But the marketing and sales guys, who don't give a shit about ease of development, robustness and so on, care a lot. Since they get pressure and promises from giant companies (IBM, etc.), and since they don't want to get blamed due to bad sales, they just push for WebSphere or BEA, thinking that the client will be glad. Of course, when IT asks for the reasons of the Websphere choice, they say that the end clients want Websphere, and that it will help selling the product...
Guglielmo Lichtner spake:
> However, I sincerely doubt that BEA will be packing it in
> next year, as Marc claims.
> The reason lies in the psychology of IT managers.
A disfunctional lot, as we all know. I'm afraid I have to agree with you here, but:
> What they do care about is features that allow them to keep
> their jobs, and those features are 1) reliability, 2)
> scalability, 3) on-site support, 4) J2EE compliance.
Two points: 1) JBoss has these. 2)none of these help IT managers keep their jobs. What helps them keep their jobs is having someone else to blame. If they can blame IBM for having a crap product, then it's not their fault, somehow, for choosing the crap product. If they choose JBoss (or postgresql, for example) they're responsible for making it work. What I like is being _empowered_ to make it work. But I'm not a manager, I'm an architect. I don't like blaming people, I like making things work.
>> What they do care about is features that allow them to >keep
>>their jobs, and those features are 1) reliability, 2)
>> scalability, 3) on-site support, 4) J2EE compliance.
>Two points: 1) JBoss has these. 2)none of these help IT >managers keep their jobs. What helps them keep their jobs >is having someone else to blame. If they can blame IBM for >having a crap product, then it's not their fault, somehow, >for choosing the crap product. If they choose JBoss (or >postgresql, for example) they're responsible for making it >work. What I like is being _empowered_ to make it work. >But I'm not a manager, I'm an architect. I don't like >blaming people, I like making things work.
Regarding your point (1), I think what matters is that JBoss is not *known* for being scalable. Is there a comparison of the scalability features of JBoss and WebLogic somewhere? Specifically load balancing, coherent cache and TCP/IP multiplexing. Are those all there? Are they mature? If so, I am switching over to JBoss!
Regarding your point (2). Firstly, when a company spends $3,000,000 on a project they want one thing - to serve the users. They really don't care how. Secondly, IT managers do not want somebody to blame. They do want their projects to succeed, and they have money to spend (the WebLogic license is a negligible part of my project's budget, for example) so they figure they'd better buy something that seems reasonable and mainstream. That's how people decide to buy Oracle - it's expensive but "safe".
I believe that JBoss might be a really strong J2EE server. I don't know, I have never worked on it, but for the sake of argument lets just assume that. I am a happy Weblogic user BTW.
But let me explain why I don't care about JBoss, and I will do so by making an analogy with shopping for a car.
I bought a Volvo V70 wagon a while back. I love my car. I am very happy with my decision. I also looked closely at the Passat wagon. Also a very nice car. What it came down to was I liked the total feel and features of the Volvo better. It was more expensive, but I felt that it justified the extra expense.
When shopping, I never paid more than a passing look at the engine. I compared horsepower and torque and cylinders, but that was it. I didn't read up in the journals about the technical sophistication of the engine. Of course, quality was an underlying concern, and therefore I was shopping Volvo and Volkswagon and not Hyundai. So brand was important.
My point is this, J2EE features of servers are becoming comoditized, just like automobile engines are. Yes, there are big differences in the quality and power of these components, but that is where vendor reputation and specs are important. A V6 engine from Porsche is probably a better component than a 3 cylindar engine from GM.
BUT, people don't usually buy engines. People buy cars. Even if you gave me a free Porsche V6 engine, I would still rather buy a complete car with a GM 3-cyl engine. Because I can get much more value out of a car than an engine.
If you read what BEA and Microsoft are talking about these days, you understand that they are moving to offer a complete car. Not just a container, but an integrated tool and feature set on top of the server to make application development easier.
How many developers out there have the ability to master the J2EE spec. Not many, maybe 500,000. How many VB developers are out there? Maybe 10 million? They will be able to develop to .net, they won't be able to touch JBoss. From some interviews with BEA top brass, it sounds like BEA is looking at the same math and are moving into this space.
Okay, so my final point is this: Microsoft, BEA, maybe IBM, are offering roadmaps that look far beyond just the server. They are talking about the user experience and building an integrated environment to improve that. JBoss' roadmap seems only to be concerned with the nuts and bolts, and not much more.
People want to buy a car, not engines. JBoss is trying to be the best engine on the market, but is not concerned with building a car. I will stake my career on the carmakers.
"I will do so by making an analogy with shopping for a car."
Eyes glazing over. Car salesmen analogies are up there with VHS vs. Betamax in my book.
While TCO is not an issue in your car-buying decision (just wait 'till you have to have that Volvo serviced in someplace like the US of A and you're dealing Sven's European Custom Auto Repair), it is now for most companies in this recessionary environment.
"My point is this... Even if you gave me a free Porsche V6 engine, I would still rather buy a complete car with a GM 3-cylinder engine"
JBoss is not just the engine. It's a full server. They too have webservices today. So you're saying you'd rather buy a Ford Escort than get a Porsche Carrera for free.
"How many developers out there have the ability to master the J2EE spec. Not many, maybe 500,000. How many VB developers are out there? Maybe 10 million? They will be able to develop to .net, they won't be able to touch JBoss. From some interviews with BEA top brass, it sounds like BEA is looking at the same math and are moving into this space."
So I guess the 50K people who download JBoss per month are just wanking around then? JBoss is closer in numbers to mass-market competition than BEA will ever be. No one goes against Microsoft in the mass-market and wins. No one. Open Source (Linux, Apache and JBoss?) are the only mass-market structures that can stand their ground there. Even Open Source has to prove long-term that it can continue to keep or gain market-share on Microsoft, IMHO.
Talking about .NET vs. J2EE as a spec. This a battle only Sun and Microsoft care about. All the middleware vendors will build web services on the winner. This is what Marc says in his interview.
"JBoss' roadmap seems only to be concerned with the nuts and bolts, and not much more."
Maybe you should take a closer look at their product and read Marc's interview. While I don't buy all of Marc's puffery, he and the rest of JBoss (-Rickard it seems :) aim to eat BEA's lunch. That much is clear.
Interesting analogy, although it didn't quite make sense to me.
Fortunately you are completely wrong about JBoss being all about "nuts 'n bolts". Quite the contrary. It is about making the developers life as easy as possible and also give him as much power as possible with the least hassle as possible. The features being added now, like being able to update *parts* of your app at runtime, or being able to invoke your components through SOAP/RMI/IIOP/.Net is aimed at precisely what you talk about: creating a user experience that just makes you feel good , and think "I wonder what happens if I push the pedal to the metal". And off you go :-)
Quite honestly, while Marc uses words like "take over the world" which may sound over the edge, what he is seeing is that the ultrasimplicity being brought into action in JBoss is precisely what will make JBoss attractive to not only the "nut 'n bolt" user, but also to Joe Programmer; your average VB developer. The 10 million.
It's as simple as that.
I think the real dark horse that everyone is underestimating is HP's new application server (HPAS 8.0) which oddly takes a page, er few pages, from jboss. They've built the whole thing on top of their Core Services Framework (CSF) which is actually a JSR (http://www.jcp.org/jsr/detail/111.jsp
). The CSF is runtime mgmt. framework using JNDI and JMX so services can be added/removed/managed on the fly. For HPAS 8.0 all of its services (ejb, web container, jms ...) are just components that can be added/removed/etc using JMX (they include a JMX browser) and accessed by other services using JNDI. This offers extreme extensibility and customization.
So Mark, I dig jboss and hope for the best for you and it, but I think CSF out does even the JMX incorporation/framework jboss has. HP owes you (and Oberg) a thanks for the ideas.
Guglielmo said, "What they do care about is features that allow them to keep their jobs".
That is true. Although I hope that his way of thinking will change. May be I am an IT poet, dreaming about the ideal IT manager, but I will definitely not belong to this kind.
You also said "JBoss should stop refactoring its architecture and start concentrating on load-balancing, distributed caching".
I think this is the purpose of the Rabbit-hole project: to develop a high quality application server with all the "Enterprise Features" for free. And they will get there. No doubt about it.
More and more commercial applications will support jBoss. More and more companies will provide technical expertise and support for jBoss. More and more companies will provide training for jBoss. That is obvious. But it is worth mentioning it.
As an independent consultant with expertise in BEA, WebSphere and Orion, I was happily surprised when I started to receive calls from customers to work on jBoss projects…
i agree with you Darryl we develop an application framework based on ideas from HP and JBoss and i wish that soon JSR 111 become a standard.
So the main difference for us between CSF & JBoss is an extensive lifecycle for the app. services and a more difficult way to write new services, we use the lifecycle from JBoss wich is a lot more simple and powerfull.
So we are moving now to add more app. services like workflow and personalization and i´m interested in share some ideas with you and the community. (jcabrera at cubika dot com)
I also have a question for Marc Fleury: Why is it so difficult to find out about the JBoss support matrix for EJB, JMS, etc? Should it not be on the front page? For example, I gave up trying to find out whether Jboss supports container managed relationships.