je ne comprende pas
One message from today's announcement: Orion is dead. The much ballyhoued initiative from JS, that apparently warranted his rise to President has been killed, gone are subscriptions for software licenses, in favor of enterprise contracts with Sun Services. I guess his boys on Wall St. who were claiming that most of Sun's profits would come from software can't quite keep up with its all-aware COO. Now, SLA's are the only thing left of iPlanet. What a shame - - what a joke. Making AS 8.1, EE free is yet another nod toward BEA's business. I will take official credit on this Blog for making Sun's app server free, and I give PP credit for making it OSS. The only thing that could have made this announcement anything but yet another non-release of info. would have been to describe Glassfish indemnification. Are we to presume Sun will do this? Or will that hurt WebLogic too much? As far as making Solaris a platform of choice, why not do reference architectures, (see first link in my post, here:
http://douglasdooley.blogspot.com/2005/11/step-2.htmlto demonstrate the integration of the platform. It just makes me look forward to better days everytime JS gets on stage and proves another one of his volume theories with vaporware. Even MSFT has web-sites dedicated to software integration, how can Sun not? Let me be clear to those in Sun and those outside supporting it, and those external on life support because of continued non-moves by the supposed-to-be software leader, the days (more likely, years, unfortunately) of inaction are numbered. Whether it is me, Marc, or some girl from Beijing, there will be a change that will sweep away possible plays of inaction by Sun, and will lead the charge through the development community to a new mantra of determination (more to come), based on skills and not politics. I am tired of hearing nothing coming out, except from MB, about the strongest portfolio in the market. I am tired of gearing up for something big, and having another let-down. It's already time for a change. It will just take longer than anticipated back in 1999, but make no mistake JL, JS, and all the infrastructure that supports the current software administration, we shall be back, and this time we'll be stronger than you...
announcement
Looks like we have a re-positioning of the subscription pricing model coming tomorrow, unless there is something to be said about the Desktop program at Sun. Maybe Star Office 8 needs to be re-launched or something equally lame. In any event, unless MB is involved, it is bound to be completely lame, especially considering that the current barometer is the open source database conversation - - who cares? Well, maybe Solaris developers care that Sun is giving away an already free database, but when it comes to Java, database support is a given. I may care to retract my post, if JS and JL announce something like Glassfish integration with JAX, but that would require JJ being on board. Maybe they will announce that AS 8.1, EE or AS9, PE is going to be integrated in to N1 instances and/or Solaris 10. In any event, it seems unlikely that there will be much movement towards policies, practices, and products that actually enhance the Sun value proposition vis-a-vis, umm, maybe BEA? I am so sick of the garbage coming out of that company in the hopes that WebLogic will succeed. It's over, guys, read my posts, its obvious we won this battle, and have therefore wrestled away from you the gameplan. You are now playing by JBoss' rules, and you have the developers to thank for that.
So come up with something bold, and not just announce something that everyone will groan over. You are Sun, you're supposed to be leading an evolution so vast that it changes how business is done. If it has anything to do with re-pricing, re-announcing, or re-positioning, you have failed once again, and you might as well cancel the press release. If it is just to upstage Dec. 6, fine, its a business move, so invisibly veiled. But if you have something that MB and JJ can support, then you might get some support from the constituency that you are supposed to be feeding: the JEE developer community. Until there is some recognition of the state of affairs in terms of app server marketing, there will be no hope of a resurgence of Sun's business. I can foresee more great announcements every J1 from MB, especially next year with an integrated Glassfish w/JBI (if that happens next year). But until the words "fail-over" grace the lips of JS, there will be an also-ran strategy, and a future strategy. Are you really going to leave it to Marc to continue to dictate terms? Are you really going to leave the major movement to me?
JEE 5
Like JBoss, AS7, EJB 2.0, JAX-RPC, AS8.1, EE, Glassfish, and the incremental steps that have come before, the Java community finally has the platform for the future. I don't know what the coders are talking about with POJO, other than assuming J2SE is easier than J2EE has been, but I do know, judging from my man on the inside, that EJB 3.0 is significant and worthy of an entirely new release of the Java Enterprise Platform. Java Enterprise Edition 5 has been released:
http://jcp.org/en/jsr/detail?id=244(sorry for the redundancy, tss.com)
To make developing EJBs a little less like major surgery, the expert team and all the contributors have done an admirable job of adding annotations, persistence, and web services support in order to achieve a true enterprise technology. EJBs are to the Internet what ODBC was to the Client-Server model, and if you don't believe me, than you're running a counter campaign, or just not paying much credence to this blog or JBoss' business. When you combine portability with scalability, you have the makings of technology to run with for the foreseeable future.
As I indicated in 'Java Implementations 1/2/4':
http://douglasdooley.blogspot.com/2005/08/java-implementations-124.htmlpre-built components are the gift of the Java Platform, Enterprise Edition. And in Java Economics, I indidcated that there is one technology to know in the web services world, and that is JAX:
http://douglasdooley.blogspot.com/2005/07/java-economics.htmlSo those are the technologies, and we have spoken long about the need to rally around these two, as well as JBI:
http://douglasdooley.blogspot.com/2005/11/jbi.htmlBut I would like to have a more focused discussion about the impact that this will have on Sun's business, and what it means for the Java development community. From Sun's vantage point, I hear nothing. JS would rather talk up Open Solaris, which is the former crown jewel, as opposed to the new crown jewel in JEE 5. I understand his motivations, and I guess I kind of think that it is worth the effort to keep a Sun OS around. But that is in a vacuum, and the reality is more likely to be that the OS investment will come at the expense of the Java OS investment. That is not necessarily a huge deal at this very point in time, but come about the time when Glassfish and Clustra need to integrate together, where will Sun be? They will be talking up the 4 millionth download of Open Solaris (correction to IT media, that does not mean 4 million paid licenses). But they won't be earning a damn cent from all this P.R. In the meantime, JBoss is happily building a platform via Hibernate and JMX. I don't mind, I'll probably be aligned with the latter after creating the former, but it goes to show that the Sun shareholders don't know anything, I mean nothing, about software. They think Orion, Liberty, and Net Beans amount to a strategy - - all three are in ruins, and the company continues to go adrift. Like the talk around MI, there needs to be a clean house, but of course that won't come.
Rather, there will be a continuation of a broken strategy based around Solaris, while the Java Platform becomes a viral marketplace. That is where the developers come in. We started off on a mission to deliver Java web services out of pre-built functionality. This began in September of 2000 with the release of EJB components from Diamelle:
http://www.diamelle.com/news/iPlanet.htmdesigned to run with any compliant app server, iPlanet being the initial target. At or around this time, BEA was trying to develop the Theory Ceneter components into pre-packaged commerce applications. I recall the TC components being a major driver of WebLogic customer adoption, with their ability to demonstrate the implementation of J2EE with real-world applications, something that Sun and IBM were slow to demonstrate (San Francisco was not J2EE). When Diamelle released portable components, not tied to any app server platform, the value proposition of J2EE became more evident. I would argue that it was the apps - - the J2EE components - - did more to promote the J2EE brand than the total number of app server platforms. After all, were not the app server platforms all chasing the same goal of customer adoption, while the component implementations were actually demonstrating what J2EE is?
I wrote that last paragraph a few years ago, maybe there is more to come...
What are the developers to do? Probably stop working on all of these open source projects, including Portals, App servers, and Databases, and let JBoss get their due, while the real money is to be made in the apps. What I mean by this, is that it is time to start writing EE products that work for ME devices, and leave the infrastructure to the current players. The race is over JBoss won, and now they will have to fight for viability with Glassfish, and to a lesser extent Geronimo and WebLogic. But there is an unending opportunity to build the services and the components that will develop into services, while there is PS organizations writing nothing other than propreitary code. Its a shame. All this talent out there worried about Marc's ambitions, when he is building a neutral platform (if you think Geronimo is neutral, I have some work to do) for all of you to write JWS. I come at you with the lyrical, spiritual, raw-sh_t I spit at you...
step 2
deploying JWS, this time update with UltraSparc T1 servers and Glassfish:
(warning: this is a .pdf)
http://www.sun.com/service/refarch/collateral/WebServices_Bro_V1.pdfJava application servers, who would have thought a company would build a chip architecture based on them. It took 10 years to develop Java, 5 years to establish J2EE, and now there is little need to think about Java on a chip, for it is already built in to the objective, which is multi-threaded, multi-instance performance with the fail-over at the software level. It would be a mistake to think of this as an alternative to Opteron, for this is not a web server strategy (or grid if you believe the marketing:
http://www.sun.com/servers/dynamicweb/), this is a Java strategy, straight-up. There is nothing like it in the market, even though Azul is doing something interesting, though its viability is somewhat up in the air. Java processing on the hardware, now competition has a new level of complexity, how does BEA stay in the app server business. If anyone has any doubt that we did the right thing on 10/28/02 (
http://douglasdooley.blogspot.com/2005/10/102802-102805.html), this should put it all to rest. What if Sun reps. still had to decide which app server to sell on the Java hardware, would that not make the experience of 1999-2001 seem minor in comparison. Would there not be less cohesion in the enterprise company's offering? I contend that the largest single product release in the company's Internet history was the Sun ONE Application Server 7. If not, BEA would be part of Sun - - which name would the company be known by? Who would be its leadership? What kind of IT market would we have?
This Open Solaris stuff is cute. I can see JS' point of view that in order to stay relevant, it needs to adapt. But Sun has been good at software functionality, though it has not been successful in Net Beans. It has been so thoroughly beaten in IDE platforms, that people are starting to talk about keeping them around for a safety #2. And what in the heck is going on with Liberty? Has JS started anything that has succeeded, including Project Orion. The sale of the app server via subscription model is an abject failure. Singing up GM does not compete with WebSphere, let alone displace WebLogic on Sun boxes. The amount of F.U.D. that has been spread on Sun's app server is the most of any product in the history of software. This includes the internal discrediting by nearly the entire Java team under JS, as well as the Java tools team (DR, you're guilty). What a crew of wannabe product managers, who have never really been in charge of a real product.
The only product that they have been in charge of that has any consequence on Sun's business is WebLogic by proxy of their treason. I know, big words, but when you try and run a product in to the ground, for reasons that are attributal to another company, you are out of bounds. I can only say that we (and that is a small we, because it was a minority that overcame) placed Sun in a position to turn-around today without resorting to some AquaLogic positioning. Scott should be giving props to DH, PP, RV, DB, RL, and DD, and I am sorry to others, but that is about it. Today, he can walk in to a customer, say that he has the best chip in the industry, best hardware again, and for the first time, a competitive, market-changing, systems vendors' top app server. What he has been waiting for, for ten years is real and here.
(BTW, Jonathan: Scott announced this in '03, there's nothing new here:
http://www.sun.com/smi/Press/sunflash/2005-11/sunflash.20051115.4.htmlhttp://blogs.sun.com/roller/page/jonathan?entry=rebuilding_a_great_partnership )
I know he is the next CEO, but for reasons that only some can follow, he is not the man to bring about the transformation of an industry, and an economy. He is only there to fill space. I credit him with being there, and that is it. Java was started before him, app server was created in spite of him, Liberty Net Beans, and the Java Studio line have been failures, can we expect more from Open Solaris? I am not confident, the OS is dead, JWS lives, that is something that he fails to grasp. I am sorry, JS, but your years are numbered even before they start, be prepared becuase the tide will roll-in...
JBI
In case some of you punters out there did not get a good analysis on the Internet Services (not to be confused with Web services) discussion flowing out of MSFT, I will provide a link to a must see article, especially for Friday:
http://www.theregister.com/2005/11/10/gates_ozzie_memos/On the flip-side, it looks like my old colleague has been promoted, makes one wonder:
http://www.infoworld.com/article/05/11/03/HNjavanet_1.html?WEB%20SERVICESI guess everyone now knows SOA. It's the - - in the words of Red Herring - - industry pixy dust, just say it, sprinkle a little over any old product, and voila, you have the key to integration. Ahhh, I'll just keep movin on...
What I came to discuss today was the JBI spec. And in the interest of helping out my old big boss, I will do some marketing for it, though I admit to be a newcomer to SE and BC, but that's what the spec is for. This will be in hopes of coming one step closer to an understanding and an agreement that standardized components is the only mantra to know. I am admittedly shy about the liberal use of the word components in JBI, especially when an app server is considered a component. But its not the fault of the JBI Expert Group that no one has come close with pre-built components, made out of the smaller variety and more specified functionality. In time, components will be more akin to composite services than to service providers. I am sure there is a roadmap, but where exactly does the service run, can an EJB expose itself via JAX-RPC and be built for a composite service in the Service Engine of JBI?
It's really a topic for the integration players in the market, and with SeeBeyond gone, there aren't that many integration players, at least if you are thinking in current models. Once JBoss comes out with an ESB, they will place even more fear in a market already running for cover (even my favorite integration vendor's CEO has misplaced mis-givings, and couples that with acknowledgement for BEA's and IBM's abstentions:
http://www.capeclear.com/annrai/archives/2005/06/one_final_note.html)
What is a composite application? What would a SOA web service look like? We did a web service example for a whitepaper some time back on "EJB Components: The Building Blocks of Web Services." Some of you might have seen it posted in Java World (where is the love for the effort?) back in '01 and '02. In that paper, we had a credit check example with asynchornous messaging, and something else that I forget. The key was that it was a series of Session Beans exposed as services. So my question for Annrai is, what exactly do you write software in before it becomes a service? Or are you supposed to use your very WebLogic Workshop-like Cape Studio tool just to build XML files. It's got to be written in something, unless we are convinced that all the possible services scenarios are already written and all we need is EAI-like qualities to expose them. I, for one, doubt it.
You're going to either use VB or Java. End of story, everything else is dead. All of these punters, PY included, running around claiming Java is dead are self-interested in this cause, explaining that LAMP and AJAX do more, or are better aligned, or whatever they claim. Java, by virtue of MB and JBoss is here to stay. BEA can take their Aqua Logic and spin it all they want (BR, get a haircut), but the fact remains that software developers, whether on the SOA tier or not, have to write in something. The services will be composite of many things, including new components. Don't make me go into a diatribe on component-development as a means to services implementation. I leave the diatribes to Marc.
I wonder what Apache is going to do, as they fork their commitment of J2EE and JBI. Will that leave them competitively stronger or weaker? I believe they will go stronger based on Geronimo and IBM, which is pretty obvious. But I think they just leave a whole lot on the table for JBoss to clean up with their inability to innovate on the JBI front. That leaves Glassfish and JBoss as the two remaining app servers with the complete integration solution. The only competition to them is a WebLogic/Oracle alliance. But it looks like that would take some time.
I have read the JBI spec, not really in its entirety, but enough to get some of the drift of the SE and BC (not really). I, of course, applaud the efforts. I wonder what the JAX team thinks, and how they will be integrating. Not really sure about this J2SE craziness, but I'll give them a few years. Customers are doing SOA, but it shouldn't take so long to get up and running, so the vendors should come ready to deliver in '06. JBoss customer should immediately adopt Messaging, and Sun customers should be working with the JBI RI. Anything less is a delay to the inevitable. I don't use that term lightly, it is much too important for businesses that have made the correct decision to go with SJ AS 8.1 and JBoss 4 to get ready for JEE 5, Glassfish, and JBI. I may sound repetitive but that is not just for emphasis, it's just because no one else is talking about it...
I still can't believe Ray Ozzie's blog, or whatever you want to call that message to executive staff, which became the Internet public at large. But even though El Reg has comically pointed out its effusiveness, I think it did something else: it upstaged the product launch. Something tells me that is exactly its intention. Look, companies compete internally as well as externally. And SQL Server is a maintenance business while the Ozzie memo is a different kind of services model. I can see the logic that they are pursuing, but I still wonder why kill Hailstorm, if you're going to do it with consumers? Is it anti-trust or is it the possibility that they feel they would lose a comparative Java web services battle? I kind of think the latter, knowing how they operate over the years. The Hailstorm initiative, which I have written about:
http://douglasdooley.blogspot.com/2005/06/components.htmlis a logical next step for MSFT under Ozzie. Who is calling the shots? Back to the silence, no more dropping science, everybody rapping about diamonds and violence...
JBI contains internal: SE and external:BC. Software "engineers" unite, as there is more to come...
the difference
With JavaOne Tokyo and the Microsoft launches, there is a lot going on in November already.
One thing is for sure, adaptors are proprietary:
http://www.theregister.com/2005/11/07/biztalk_adaptors/this includes you, Sun, and your iWay fiasco...
And what is the deal with Gates flying around getting all upset that people can access things to do on the Internet:
http://www.theregister.com/2005/11/09/gates_warns_disruption_to_come/Unless he is talking about .Net, he needs to stay on message regarding B2C v. B2B focus. B2C is something for the media companies, while MSFT is a software company. I can rationalize the move to Live, but cannot understand the obsession with Google. Ray Ozzie is not on my level, if advertising revenue is his model:
http://www.scripting.com/disruption/ozzie/TheInternetServicesDisruptio.htmCharge for transactions, Ray, businesses will pay. There is only one way to be looking at the future of software, and one of the key steps is found here:
http://www.sun.com/smi/Press/sunflash/2005-11/sunflash.20051104.1.htmlmore to come on this issue...
free for all
