Really? IBM software to run the same apps,without recompiling on MAC OS X and Linux..., but wait a minute, wasn't that the job of JAVA? Remember? "Compile once, run everywhere?", that turned out to be "Run once, compile everywhere?"...

Ah. Mainframes dont crash. Midrange unix boxes (pSeries, Solaris, HPUX - you name it), dont crash. iSeries - whilst out of the Ark - doesnt crash.

Example. One of my customers run a 10,000 user Lotus Notes site, running on zSeries (IBM 360 class monster mainframe). "When was the last time the server crashed? ". "Ah. Never."

Ironically, MS have locked their application suites into Wintel, but dont reliably offer any of the clustering and replication services thats required to actually provide 24x7 service. Take Exchange, for example...

And MS are surprised that business want to host enterprise-class applications on reliable hardware ?

The Cost example is also a false one. Less servers == less management overhead and leads to less cost. Instead of running a single exchange server (on a machine, etc) for every 500 or so mailboxes (due to shared store limitations and risk), run 10,000 users on a single machine.

Perhaps this is why they're finding it difficult to get into the enterprise data center for mission critical apps at the top end of their market, and finding zero-cost solutions such as Linux eating their lunch in the lower end.

---* Bill

Evan, your post just exhibits your ignorance. First of all you can have two binaries. Not much problem there. A lot of Linux distros support x86 and Mac... errr Mac is x86 mhhh. There's a thought for you!

Now if Apple (a small thing compared to IBM) managed to get Universal binaries out the door and have apps run on PowerPC and x86 (the same app). What do you think IBM can achieve? Its partners?

Finally, the days of Java's problems are gone. BTW, the joke was "Write once, debug everywhere" not "Run once, compile everywhere". What are you some sort of open source gcc junkie???

Well getting back on track after that laughable moment. Java now runs well on all platforms with no recompile and given todays tools you can configure OS specific dependencies without hardcoding them. Not to mention the fact that a great deal of Java usage is on rich web applications which require only a browser.

So next time you pull a post like that at least get the joke right.

"...but dont reliably offer any of the clustering and replication services thats required to actually provide 24x7 service. Take Exchange, for example..."

I find it very funny that people who portray themselves as knowledgeable about computing do so little research when they decide to spout off on public forums. Everything Wild Bill says about Exchange was true for v5.5 but stopped being relevant 4-5 years ago. I suspect that that's when he had experience with it. Microsoft Cluster Service (MSCS) has been around for a long while now and there are plenty of 3rd parties supporting synchronous and asynchronous replication (HDS has a fine product that leverages MSCS and MPIO).

Microsoft themselves have included LCR and CCR replication in v2007 (look it up...for smaller installations) and people have been running mail-stores of 3-5000 users per box for some time as well.

Anyone who espouses Notes in this day and age is hopelessly clueless simply based on it's performance characteristics and it's gargantuan resource requirements.

For what it's worth, Trevor... I think you just did the same thing you accused Bill of...

Does Hilf have a point or is Microsoft just trying to blunt the possible effect this will have on its desktop monopoly?

Mostly the latter.

Bill Hilf needs to just keep his mouth shut. He does nothing but spread FUD ever since he left IBM and joined Microsoft.

I agree with Bill. I worked with exchange servers for many years, and whilst there is nothing wrong with them, i am quite sure there isnt any version of exchange that will even come close to main frame reliability. Besides, running many servers is a management nightmare. I like Windows, but I cant see how one can claim that Windows is not management intensive.

What do you expect from Bill Hilf? I mean really, are some people still that naive?

Here's a couple of quotes from Bill Hilfs mouth:
"At the end of the day, we're in it for business reasons,"

"I exist for business reasons. I do not exist as a PR stunt or as sort of an olive branch."

So how does anything from Bill Hilfs mouth, or any Microsoft executive for that matter, mean anything to a sane person? ANYTHING we could rely on or trust as having any ounce of truth? Pleeeease, IMO, Bill Hilf is just another Microsoft MS-SnakeOil salesman.

quotes from: http://articles.techrepublic.com.com/2100-3513-5829512.html

I'm happy anytime MS gets upset at the competition in the marketplace. It means that the competition is doing something right. I have worked in a mainframe shop in the past and I now work in a MS shop. Both have their advantages and disadvantages, but overall the mainframe environment is extremely stable in comparison to anything that MS has to offer hands down. That's why so many large companies continue to rely on the mainframe. Why break something that already works so well?

In addition, I shudder at the latest offering of Windows from MS. The OS is overpriced to what it offers. Why spend gobs of money on hardware upgrades and then the OS just to get the latest when your current platform works fine? I'll let someone else make that mistake.

Gerardo Tasistro, goto IBM and get the White Paper of how you can migrate a BEA WebLogic app to websphere and then tell me if the hundrends of pages of directions really accomplishes what java promised...The reality is, that even though java has made some advances, only a very small set of applications, too small in my opinion for any real application, can really accomplish what java promised

If Notes is so horrible, what is Ray Ozzie doing at MS?

What on Earth is proprietary about the Eclipse RCP? It's open source for goodness' sake! And the comment about "the mainframe ... is four to five orders of magnitude more expensive than anything offered on x86" is pure bunk. In large environments that need big iron and have have made the investment, utilizing spare resources on that platform to provide collaboration services just makes sense. Buying new hardware just because some executive at Microsoft says so, doesn't.

You should have defined what the Open-Client actually does, before you start asking us about it . . .

I have developed Java applications on the Windows platform and then deployed them on Unix with no code modification. I do know that there have been some problems here and there for Java, but overall, it is one of the better technologies in the IT field. To say that Java has not accomplished much since its inception is an incorrect statement. Millions of users run Java apps everyday w/o even realizing it. Heck, even my phone runs Java (the J2ME edition of course).

Evan I did and this is what came out:

http://www-128.ibm.com/developerworks/websphere/library/techarticles/0405_beaton2/0405_beaton2.html

Read here:

"It is likely that significant change will not be required in your application code as part of your migration. For the most part, application code that is "just Java" should work on both platforms. Chances are most of your control and business logic falls into this category. Even some of the J2EE artifacts like servlets, JSPs, and EJB components may work without modification on the new platform."

It then goes into explaining which things would break or change. It talks a lot about application development tools and how you'll need to change how you work. At one point it touches the topic of EJBs and proprietary technologies. Which will break code if not present.

But this issue is not a problem with Java portability it is an issue with the container. Pretty much like saying Java isn't cross platform because my app (the one that uses the serial port) doesn't run on Linux. Wohoho mister. Linux doesn't have a COM1.

You can build a proper application with the usage of interfaces that is implementation independent. In such a way that you application detects the underlying OS and uses the proper implementation. If it detects Windows it loads the class that implements the SerialComm (so to speak) interface for Windows and on Linux it loads the class that implements the same interface, but on Linux.

Am I making myself clear? The problem isn't with the language it is with the tools underlying the application's deployment.

Gerardo, you are only enforcing my point. I mentioned real -non trivial- applications. All these need an application server, and if you don't stick to the core java libraries, you are out of luck. Sticking to the core though, severily limits your options. And because I have done that in past, this article, doesn't even begin to tell the whole truth of the problems you are about to face, again on real applications. But this is not my point with IBM's announcement and maybe i implied what i wanted to say. IBM's policy of enticing new customers with open source software, to win the account, only to throw them to the super expensive mainframes, AIX, AS/400's later - when things don't quite do what you need, is not new and I no longer buy it...After all, what is this huge IBM consulting teams are going to do?

Well on the contrary I believe you made my point. My point is that it is not a language issue, but a vendor issue. And that is exactly what you said.

In any scenario if you get caught by the spin masters of your provider and buy stuff you don't really need you will end up locked in. But that is not a Java issue. It is an implementation problem. A mix between architectural design and tool selection.

If you look around you'll see lightweight frameworks emerging which provide services which were once container dependent. By using those you can free yourself of the vendor lockin in a lot of cases.

As a counter example. If you use Oracle and build your DAOs with SQL code specific for Oracle. Is Java to blame when you move to MS SQL server or DB2? I don't think so.

A bind to the vendor of the OS or a bind to the vendor of the Application server is more or less the same. Indirectly the java platform is to blame for not being able to provide these things in a vendor indepented way, bc it provides a JavaBean SPEC, just to mention one area here, vague enough to any vendor interpretation...

In Java ecosystem there are two different approach of "portability": one based on specification (eg J2EE) and another base on libraries that run everywhere (like Hiberante).
If you develop by specification your applications are really portable but there alway some things like admin-tasks that usually vendor-specific (but others specification for this task are always under development).
Otherwise, you can leverage on some great libs like Hibernate, Spring, Jakarta/Eclipse projects that are tested over multiple environments.
Mixing these two approachs you are very near WORA (better than ever other platform after all).

Hahahahaha!!!! What a foolish comment. So you intend to convince us that somehow the language is to blame for the usage one gives of it? Even if the specs are made by the same company you will be hard pressed to convince me. Take these lines from the same link

"It is very likely that your EJB code will work without modification, assuming that your beans conform to the EJB 2.0 specification. WebSphere Application Server tends to be more rigorous in its conformance to the EJB specification, so some minor violations that WebLogic Server permits may need some attention when migrating."

What they're kindly saying is that if you are to sloppy and cut corners on WebLogic you'll have problems. And that is because YOU the developer didn't conform to EJB2.0 because WebLogic wasn't rigorous enough. So to my reading it is WebLogic, not Java, which is "vague" (to use your word) in its implementation of a rather stricter standard. More so it is your fault as a developer for using those loopholes to cut corners.

Finally on the matter of EJB and the whole CMP issue it comments. Well that is clearly a problem, but given you face it day to day as your project evolves I'm sure a lot of people out there use automated tools such as XDoclet to generate such descriptors (among other things that need be synchronized). So the migration issue shouldn't be as bad as you want to make us believe and you also get some free tools to download and solve those issues.

Yet at the end of the day I couldn't have said it better than you:"IBM's policy of enticing new customers with open source software, to win the account, only to throw them to the super expensive mainframes,...". Finally it is you the consumer's responsibility to see these issues and not get caught in a vendor price-scalator.

I really think there’s something wrong with your post. I think you need to review the last part that contains your personal opinion. I don’t want to correct you but perhaps after a second look at your post, you’ll be able to realize the error.