John, first, let me say that the list would certainly benefit from these kind of comments. Perhaps you forgot to hit Reply-All?
DCOM works, it was a failure as an INTERNET service. DCOM basically is a marshaller/unmarshaller not much different from other marshallers in COM itself(the magic COM pulls when code in two different apartments communicates). It also includes _SOME_ distributed garbage collection. That's all there is to it. It's very fast, and very expensive on the network(a lotta round trips). Sounds familiar? It's very much functionally-wise like RMI (and IIOP for that matter). COM and DCOM are strongly based on DCE-RPC. I've been involved in many MTS projects and let me tell you the platform works. I've also been in 1(one) "failed" project that used DCOM(we actually reacted quickly and fixed it), and it was because we were trying to use it as an INTERNET protocol. In a sense, I was using Web Services before it was called that way(it makes a lot of sense, if you think about it when developing). Basically, many of the optimizations DCOM has make it useless in terms of it being an INTERNET protocol(for instance, try to send a big String thru the wire... It will kill both the client and the server). Nevertheless, to scale a co-located app it rocks. DCOM wasn't scraped for COM+, it just became part of the scum(SCM, the "COM Engine"). Basically, when COM+ was released along with W2K, DCOM and MTS became part of the COM Runtime, allowing for a lot of features that were missing because until then those technologies were a pseudo-hack add in for COM. The reasons we're with J2EE are: A) Less chance of vendor lock in. Super important if you reuse code across projects(a must for us). Usually, OS and RDBMS in which to run our apps is a choice we're not involved with, regrettably. B) Productivity: I can code faster with J2EE, and I don't resign control(I may delegate, but it's easy to undo that later on). C) True platform independence: as you say, enterprise runs on UNIX. That said, when I get a customer that wants things his/her way, I provide what I'm asked for. And COM does scale; in fact, DCOM allows apps to scale out much better than the J2EE app servers I've seen(penalties for physical separation of tiers are lower than J2EE's thanks to DCOM). I expect that the RMI implementations each J2EE vendor provides will become as efficient as DCOM or better in the near future, nevertheless, and from actual experience, I won't separate tiers unless absolutely necessary. I'd rather hack my way around it. Regarding COM being dead, well, THEN many of the things you "hold against" .Net will be my reasons. I honestly think, however, that Don Box meant: for enterprise class apps, COM is obsolete. I _tend_ to agree, but there are some apps that work today with COM that will not work as well without COM(to name a few: Client apps, Games, etc.) unless the replacement for COM is very very fast. That's just not the case with XML nor with the ILM. It's horribly slow. I started using SOAP because of comments Mr. Box himself did in mid 1999. It's cool, as long as it stays in an EAI role: otherwise it's bloating the apps. That's why BizTalk is a commercial failure. XML gives everybody the chance to integrate, but the parsing is shamefully slow. I also expect these to change in the near future, but on the meantime, M$ enterprise developers are stuck with COM or something slower than COM. I appreciate the chat, so I'm posting to EJB-INTEREST as well. Regards, Juan Pablo Lorandi Chief Software Architect Code Foundry Ltd. [EMAIL PROTECTED] Barberstown, Straffan, Co. Kildare, Ireland. Tel: +353-1-6012050 Fax: +353-1-6012051 Mobile: +353-86-2157900 www.codefoundry.com > -----Original Message----- > From: John Harby [mailto:jmh_inc@;hotmail.com] > Sent: Sunday, November 03, 2002 2:28 AM > To: [EMAIL PROTECTED] > Subject: RE: Huge row about J2EE vs .NET > > > I've heard from Gartner analysts before that DCOM was a > miserable failure. > This was based on their interviewing many customers who had > tried to get > something working but failed. I think reliability and overall > quality was > the main issue. I was on the CORBA side at that point and had minimal > exposure to it. Also evidence lies in that Microsoft > basically scrapped it > for COM+ and now Box also made the statement that "COM is > dead". Finally, > let's face it - enterprise runs on UNIX and above. I'll admit > to some corner > cases where someone has gotten Win to scale but I'm really > concerned with > aggregate statistics here. > > Personally, if had the money to join W3C, I'd write a > registry spec for > intranet EAI purposes if noone else has done so. You don't > really need all > of UDDI for intranet applications but the web services model > is really cool > and is far superior to the legacy integration methods. > > >From: "Juan Pablo Lorandi" <[EMAIL PROTECTED]> > >To: "'John Harby'" <[EMAIL PROTECTED]> > >Subject: RE: Huge row about J2EE vs .NET > >Date: Sat, 2 Nov 2002 19:45:55 -0000 > > > >In .Net(and in Windows DNA), there is an intranet mode, and > it's been > >there for ages: DCOM. It's lean, it's mean, and in a > controlled network > >it is extremely good performant(dare I say better than JRMP?). But > >basically, .Net short-comings come from the lack of > >declarative-anything (including services, security, fine grained > >transactions, clusterability, etc.) because they've always > been a hack > >of COM itself. Don Box has proposed the adoption of SOAP for a long > >long time, yet as you point out, it's an integration tool, nothing > >more. The fact that the transport of choice is HTTP and XML makes it > >very open(so everybody can create their own impl.) yet that is what > >makes it SO TERRIBLY SLOW. But that only means that the platform is > >different, and that there are different programming > practices. Really, > >if .Net is internet oriented, drop it and stick with Windows > DNA. But > >appart from these, I don't see any difference that would > stop me cold > >saying: "I can't make that app with M$ tools". > > > >Juan Pablo Lorandi > >Chief Software Architect > >Code Foundry Ltd. > >[EMAIL PROTECTED] > > > >Barberstown, Straffan, Co. Kildare, Ireland. > >Tel: +353-1-6012050 Fax: +353-1-6012051 > >Mobile: +353-86-2157900 > >www.codefoundry.com > > > > > > > -----Original Message----- > > > From: A mailing list for Enterprise JavaBeans development > > > [mailto:EJB-INTEREST@;JAVA.SUN.COM] On Behalf Of John Harby > > > Sent: Saturday, November 02, 2002 5:44 PM > > > To: [EMAIL PROTECTED] > > > Subject: Re: Huge row about J2EE vs .NET > > > > > > > > > When I say "weak" I mean in the area of declarative > services. I have > > > used the .Net Enterprise Architect version and it is > impressive that > > > you can drag database tables to a form and it will generate your > > > SQL, and they have rev'ed their MSMQ and MTS to support > queueing and > > > local transactions but I don't see the sort of intranet framework > > > that J2EE offers. For .Net, these are just utilities - here is > > > something if you need the database but I don't see any sort of > > > framework in terms of intranet development. > > > > > > Any transactions are just a new rev of MTS. Plus, how do > you define > > > a transaction with *internet* web services. I spoke with > Don Box at > > > the Microsoft Tech Ed conference in New Orleans and he > acknowledged > > > that internet transactions were still to be resolved. One > thing he > > > specifically mentioned was "do you want to lock your > local database > > > waiting for a response from some calls going over the internet?" > > > > > > Where you say J2EE and .Net are different, I maintain that the > > > difference lies at a higher level. The problem spaces targeted by > > > the two are separate for the most part. Why would you want to use > > > web services for some internal app unless there is some EAI > > > involved? Yet it is a real pain not to use web services > if you are > > > making calls out on the internet. Where you say: > "Regarding J2EE and > > > .NET being different, well, of course they are, but basically, > > > they're n-tiered transactional servers (some form of 2PC > > > implementation available).", I claim that the core > purpose of .Net > > > is an internet server (with transactions, security, etc. still > > > being defined). In my opinion, J2EE is focused on the > > > intranet and will support the internet while .Net is focused > > > on the internet and will support some intranet. I think this > > > difference is a kernel difference and comparisons get silly. > > > It would make more sense to develop UI and Web Services with > > > .Net that interface through to a J2EE middle/back-end using > > > the Java Web Service/.Net interoperability. > > > > > > >From: Juan Pablo Lorandi <[EMAIL PROTECTED]> > > > >Reply-To: Juan Pablo Lorandi <[EMAIL PROTECTED]> > > > >To: [EMAIL PROTECTED] > > > >Subject: Re: Huge row about J2EE vs .NET > > > >Date: Sat, 2 Nov 2002 11:57:01 -0000 > > > > > > > >How is .NET weak in the middle and back end? > > > > > > > >I tried to make a list: > > > > > > > >Transactional objects _must_ be stateless. > > > >DB must be one of: SQL Server, Oracle, Sybase(no DTC 2PC > support). > > > > > > > >But that's about it; and since many developers *prefer* it this > > > >way(they rather have everything stateless), I wonder if > it is weak. > > > >Maybe it's a matter of comfort and experience? > > > > > > > >More and more I hear that(in the scope of why J2EE failed > > > this paper): > > > > > > > >A) J2EE and .NET are different. > > > >B) Entity Beans and JDO slow down implementations. > > > > > > > >Regarding J2EE and .NET being different, well, of course > > > they are, but > > > >basically, they're n-tiered transactional servers(some > form of 2PC > > > >implementation available). They support components based > on message > > > >queues. They manage object lifetime to improve performance. > > > They both > > > >have a Web scripting language of choice to create web > pages(JSP and > > > >ASP). They both run on a VM, which potentially enables both > > > to run on > > > >different hardware/OS platforms. > > > > > > > >They're different, but differences are subtle. The fact that > > > J2EE has > > > >been till now easier for me to work with doesn't mean I > > > won't switch to > > > >.Net if required, or if it makes my life simpler. > > > > > > > >Now, the constant noise regarding Entity Beans and JDO > > > doesn't bother > > > >me anymore, but I wonder: Do all people disregard complex > > > >persistence > > > >because: > > > > > > > >They've had a bad experience in the past? (and if so, aren't > > > they open > > > >to change?) They just can't learn Entity Beans? (really? > It ain't > > > >rocket science) They just don't want to learn a new > > > technology? (kool, > > > >mo' money for me). > > > > > > > >I'd love to hear any comments on this. > > > > > > > > > > > >Juan Pablo Lorandi > > > >Chief Software Architect > > > >Code Foundry Ltd. > > > >[EMAIL PROTECTED] > > > > > > > >Barberstown, Straffan, Co. Kildare, Ireland. > > > >Tel: +353-1-6012050 Fax: +353-1-6012051 > > > >Mobile: +353-86-2157900 > > > >www.codefoundry.com > > > > > > > > > > > > > -----Original Message----- > > > > > From: John Harby [mailto:jmh_inc@;hotmail.com] > > > > > Sent: Friday, November 01, 2002 7:45 PM > > > > > To: [EMAIL PROTECTED] > > > > > Cc: [EMAIL PROTECTED] > > > > > Subject: Re: Huge row about J2EE vs .NET > > > > > > > > > > > > > > > I had the good fortune to work at HP with someone > whom you could > > > > > call the inventor of web services, Rajiv Gupta (see > > > > > link) so I have been well schooled on this argument. > > > > > (http://www.hpl.hp.com/news/2001/apr-jun/3gupta.html) > > > > > > > > > > The primary architecture of web services (as in .Net) is an > > > > > *internet* model. Almost anything that is done is an > > > internet call, > > > > > this is the point - you discover services over the > internet and > > > > > invoke them. The traditional J2EE architecture is that of an > > > > > *intranet* model, there may be a browser interface but most > > > > > calls and transactions are occuring via intranet calls. Of > > > > > course you can lump Java web services into J2EE but these > > > > > benchmarks never do that. > > > > > > > > > > The bottom line to me is that these .Net vs. J2EE comparisons > > > > > are useless - it's apples and oranges. I am obviously a big > > > fan of web > > > > > services but I really don't see much "enterprise" yet in > > > .Net. It's > > > > > typical MS, great tools, good client development but > weak in the > > > > > middle & backend. I also think that resolution of 2-phase > > > commit and > > > > > security will be > > > > > very difficult in web services - HP had a way but it was very > > > > > complex and didn't scale. I think J2EE cores with web service > > > > > interfaces make the most > > > > > sense which is the way we are going in the JCP. > > > > > > > > > > Just my .02, > > > > > John Harby > > > > > > > > > > > > > > > > > > > > >From: Juan Pablo Lorandi <[EMAIL PROTECTED]> > > > > > >Reply-To: Juan Pablo Lorandi <[EMAIL PROTECTED]> > > > > > >To: [EMAIL PROTECTED] > > > > > >Subject: Huge row about J2EE vs .NET > > > > > >Date: Fri, 1 Nov 2002 18:08:47 -0000 > > > > > > > > > > > >The original article: > > > > > > <http://www.middleware-company.com/j2eedotnetbench/> > > > > > >http://www.middleware-company.com/j2eedotnetbench/ > > > > > > > > > > > >The analisys on the benchmark by Rickard Oberg. > > > > > > <http://dreambean.com/petstore.html> > > > > > >http://dreambean.com/petstore.html > > > > > > > > > > > >Comments? > > > > > > > > > > > >Juan Pablo Lorandi > > > > > >Chief Software Architect > > > > > >Code Foundry Ltd. > > > > > >[EMAIL PROTECTED] > > > > > > > > > > > >Barberstown, Straffan, Co. Kildare, Ireland. > > > > > >Tel: +353-1-6012050 Fax: +353-1-6012051 > > > > > >Mobile: +353-86-2157900 > > > > > >www.codefoundry.com <http://www.codefoundry.com/> > > > > > > > > > > > >Disclaimer: > > > > > > > > > > > >Opinions expressed are entirely personal and bear no > > > relevance to > > > > > >opinions held by my employer. Code Foundry Ltd.'s > > > opinion is that I > > > > > >should get back to work. > > > > > > > > > > > > > > > > > > > > > > ________________________________________________________________ > > > > > _ > > > > > Get faster connections -- switch to MSN Internet Access! > > > > > http://resourcecenter.msn.com/access/plans/default.asp > > > > > > > > > > > > > > > > > >============================================================= > > > ========== > > > >===To > > > >unsubscribe, send email to [EMAIL PROTECTED] and include > > > in the body > > > >of the message "signoff EJB-INTEREST". For general help, > > > send email to > > > >[EMAIL PROTECTED] and include in the body of the message > > > >"help". > > > > > > > > > _________________________________________________________________ > > > Get faster connections -- switch to MSN Internet Access! > > > http://resourcecenter.msn.com/access/plans/def> ault.asp > > > > > > > > > > > > ============================================================== > > > ============= > > > To unsubscribe, send email to [EMAIL PROTECTED] and > include in > > > the body of the message "signoff EJB-INTEREST". For general help, > > > send email to [EMAIL PROTECTED] and include in the > body of the > > > message "help". > > > > > > > > > _________________________________________________________________ > Unlimited Internet access -- and 2 months free!� Try MSN. > http://resourcecenter.msn.com/access/plans/2monthsfree.asp > > ==========================================================================To unsubscribe, send email to [EMAIL PROTECTED] and include in the body of the message "signoff EJB-INTEREST". For general help, send email to [EMAIL PROTECTED] and include in the body of the message "help".
