.88888 - The Fabled "Five Eights" of J2EE Availability
I was going to write about this last week, but the simple truth is that I didn't have the heart to send such a dark and somber message just before the American Thanksgiving holiday. It wasn't like the problem was going to magically be solved, and I just didn't think the negative focus would do much good as prelude to one of my favorite holidays.

Well, the turkey is cooked and eaten, the holiday is over, and many of us now have a few extra pounds to work off in the gym. It's time for things to get back to normal in the "harsher than reality television" world of Java technology that we all enjoy so well. Today's episode, my friends, is about the stunning performance (or lack thereof) reported in Wily Technology's recently completed " Benchmark Survey of J2EE Application Performance and Availability. "

First, since the results are less than glowing, I want to address the questions of methodology which inevitably arise whenever someone does a survey that produces results someone else is unhappy about. The people at Wily Technology appear to have made a responsible effort to gather useful information from a reasonable sample set, and they seem to have been even-handed in their interpretation of the data. No survey is perfect, but if we choose to disregard this one we are just burying our heads in the sand. As far as I can tell, this survey is reporting the sad but true situation that our customers and colleagues are having to endure with J2EE applications and infrastructure.

The bottom line is that way too many of the J2EE applications deployed in enterprises today have absolutely dismal performance. To put it bluntly, our applications stink! You can completely forget all the haughty talk about "five nines" reliability. The average availability of the J2EE applications referenced in the survey is just 88%. That's right - we're talking the fabled "five eights" of J2EE availability. Half of the responses show overall app availability (an application's ability to service user requests) below 96%, which means almost seven hours of downtime per week! That's seven hours of pain, lost revenue, confusion, and bad news for Java. Not only that, but half the responses are indicating even less than 96% availability.

Wily put it plainly, "Is 24x7 Availability A Reality?...No, it's more like 24x6." Believe me, I take no joy from reporting this poor performance statistic, but it's a very serious signal that we need to do a lot better if we want to hold our heads up proudly.

Make no mistake, however. Wily is absolutely NOT saying that Java and J2EE are not ready for prime time - they are Java advocates just like we are! Instead, they are making it clear that prevailing practices for designing, deploying, maintaining and debugging these J2EE applications need to change radically to favor better performance and availability for our customers. They are urging us to look at the problem from a broad perspective, too. There's a lot more to it than mere application design and development.

In fact, very frequently the problem is not in the actual coding of the J2EE applications. The cause of an application problem is a code problem only one time out of eight. The environment and connected systems are often the sources of performance issues. The complexity of maintaining J2EE applications can be overwhelming, especially when they are running in a distributed context on heterogeneous hardware and operating systems. Your company's resident J2EE "genius" may have personally designed and coded the application, but that doesn't mean that Mr. Genius will be standing by monitoring its performance every day. The person whose job it is to keep the systems running may not have a clue what "container managed persistence" is, but that's the person who is often sitting at the admin console where signs of impending failure should first be detected.

Speaking of admin consoles, it doesn't appear that fancy, expensive tools like they have in "Star Trek: TNG" will do you much good, either. On average only 42% of apps perform as planned on their initial deployment, and using a traditional systems management tool makes little difference to the results. We need more specialized tools to pinpoint and fix problems in J2EE applications than the enterprise formerly needed to keep its CICS system running.

Problem detection is a serious issue. In 40% of cases first notice of application problems are customer or executive complaints. You'll never guess what happens most frequently after the angry executive has called in to complain about the problem? The most common first escalation reaction is to call a group meeting. It suggests visions of all the people in " Office Space " having a meeting because the accounting application at Penetrode is failing and Lumbergh has just called to say, " Ahhhhh, we have sort of a problem here. " I suppose "CYA" is a necessary evil in the corporate world, but it doesn't often get any problems fixed.

After the group meeting, actual problem diagnosis takes too long. It takes more than eight hours on average to get an accurate diagnosis, and more than a full day for 30% of organizations. It reminds me of the infamous power grid failure that blacked out New York City and most of the Northeastern USA this past August. Far too often the support teams responsible for keeping J2EE applications running simply cannot identify the cause of a failure. Tools and processes which isolate problems more rapidly are needed.

It would be my preference to point out the silver lining in all of this, but I don't see a silver lining here. I believe the folks at Wily Technology have done a genuine service to our Java community, even though the findings of their survey are stark and disturbing. We all need to do a better job, from start to finish, in the process of building, deploying and maintaining high performance J2EE applications. "Five eights" won't cut the mustard, it's just plain lousy.

Wily Technology is conducting a series of free seminars in 9 North American cities right now to present more in-depth analysis of the results of this study. If one is close enough to you, then you might want to attend and get deeper insight into the problem and the solutions Wily recommends.

Please click here to join me in a discussion of the Wily survey and your experiences with overall J2EE application performance and availability. I'll look forward to hearing your thoughts and reflections.

Until next week,
Rick Ross
rick@javalobby.org
AIM or Yahoo Messenger: RickRossJL

PS - I am working on getting permission to share the complete Wily Technology report with the Javalobby membership. If I can, then I will email you a download link later this week.

VM Sharing – Could It Be True?
Yes that's right, you heard it here first. Javalobby has the exclusive announcement from one of its members who noticed on the BugParade that an essential bug focused on VM sharing in the Java JDK has been fixed! Last night we saw the announcement cross Javalobby that indeed a portion of VM sharing has been implemented in the latest internal Tiger (JDK 1.5) build. All of the Swing proponents have been sacrificing the proper animals each night in the hopes that Sun engineers will burn the midnight oil and complete the much needed features that could just make Java on the client a reality. With this VM sharing feature and the announcements over the summer that the top PC makers will be distributing an up-to-date JDK with their machines, I believe that we'll begin to see more and more client Java applications as developers get their mitts on the new JDK early next year. This seems like quite a Christmas present, especially for the people participating in the new alpha program that Javalobby is helping Sun organize. I for one cannot wait to get my hands on that build and take it for a spin around the block! Of course, enthusiasm is warranted, but as we all know there is the chance that VM sharing will not make it into JDK 1.5 and there are still a few outstanding bugs surrounding true VM sharing that could put a damper on everyone's day. As always, we'll be waiting eagerly to see what Sun can deliver in our stockings this Christmas season.

Is VM Sharing All That Important?
So how important really is this VM sharing stuff that everyone keeps talking about? Apple has had some form of VM sharing in their JDK since 1.3 I believe, and you can certainly tell a difference as a Java application runs almost as fast a native application. On the other hand, as machines get faster and faster, such as one of our development machines in the office, I believe I'll have a hard time telling whether the VM sharing makes an major improvements. On that machine, everything starts up in under a second, so its hard to judge! Where I imagine the most improvement will come is in the mid-range machines with mid-range levels of RAM. On these machines, it is important that each Java application doesn't waste 25MB of RAM each time it starts. As it turns out, that works in the favor of most developers on the client side – most of our clients will be using these machines. As much as Moore 's Law is touted, most consumers lag way behind the increase in speed that most developers crave for. So to sum it up, is how important is VM sharing – VERY – if done properly!

So What's On Your List?
Each year as Christmas rolls around, the geek inside all of us starts to get the wants for the latest and greatest high-tech gadgets. This year promises to be no different, especially for me and its getting hard to choose. Last year I treated myself to a new 15” 1Ghz Powerbook, so what shall I find this year? Browsing around Cnet.com, I found something that should wow the person who loves movies – the i-O Display Systems i-glasses HRV Pro . These goggles let you have your own personal 70” screen without disturbing anyone else. If you already have a cool sound system in your house, then I don't imagine there will be anything finer than laying on the couch with these things strapped to your head. The downside? These things are very expensive. Now if only someone could donate a set to me! So what else is on my list you ask? My friend recently mentioned to me that Tivo was preparing the next generation of their DVR that contains a DVD-R along with the internal hard drive for recording. Now, I don't think that this new series will be ready for the Christmas season since I can't find anything on their website about it, but being able to record your favorite movies or series off the digital cable or satellite would totally rock! I think that this gadget is one thing I might need to buy when it comes out. Finally, I've been having the urge to start burning things to DVDs instead of these piddly 700MB CD-Rs. To rectify this situation, perhaps I'll get a Digital Research Internal DVD±R/±RW/CD-RW Drive . This puppy burns some DVD formats at as high as 8x. Not to mention burning CDs as well. Quite a package and the price isn't that bad either. If anyone wants to donate any of these gifts to me, please email me at matt@javalobby.org.

'Till Next Time,
Matthew Schmidt
matt@javalobby.org

Sun co-founder and Java champion, Bill Joy, gave this interview to Wired shortly after leaving Sun. It has a few items which may interest us. I'm curious how many of us agree/disagree with him on issues like Jini, Windows, Apple, the future, and of course, Java. Here are a few excerpts:

(...)No doubt, but you were pretty loosely tethered at the best of times.
I've always said that all successful systems were small systems initially. Great, world-changing things - Java, for instance - always start small. The ideal project is one where people don't have meetings, they have lunch. The size of the team should be the size of the lunch table.

More excerpts in the full JL posting, please read on...

I use Swing for java programing, so I was really excited, when I saw the free, open source SwingWT library. There is a lot of pro and contra comments about Swing and SWT, but this library can connect those two worlds. With this library, you can program in Swing, make jars and build normal Swing application, then preprocess source code, and compile into SWT application. Here is short description of the library:

SwingWT: "100% pure Java library which very closely resembles the interface of Swing. The difference is that instead of using the Swing library, it drives native peer widgets from SWT (the Eclipse GUI toolkit). With this library, Java/Swing applications can be compiled natively under Linux using gcj. It also allows Swing apps to use native widgets."

Java is starting to enter a phase where there is an increased desire to create desktop applications with Java. Many recent developments such as SWT have sparked renewed interest in building desktop applications with Java.

While the underlying Windowing Toolkits such as Swing and SWT allow for the development of widgets and component libraries, outside of JIDE Software - there has been a surprising lack of high quality 3rd party custom components being heavily marketed to the Java GUI development community. If there are others, where are they? Who's making them? How do they perform? Why aren't they more 'vocal' about their existence?

All-Java MP3 Player With 1-Meg WebStart Download
I admit that I'm a sucker for Java desktop applications, but this all-Java MP3 player by JL member Jens Hohl seems really noteworthy. "JamP" is a Java WinAmp clone with skins, equalizer, playlist window and a plugin interface. You can simply click to try the Java WebStart version, or download it and run it as a local app.

It probably goes without saying some people will respond with, "it lacks feature X" and "it's not as small as my favorite native-code player." I'll just ask you to try to see the glass as more than half-full rather than a little bit empty. JamP is undoubtedly one of the most directly accessible, hit them square between the eyes, consumer-focused Java applications I have seen in a while. A 1-meg WebStart download and it's running! I'd love to see this one perfected and a lot more applications like this!

The shared Java VM bug has been fixed! I had voted for the bug and got the notification yesterday that it had been fixed. Looks like it'll make it into Tiger. This is a huge enabler for Java GUI applications, especially little ones!

Here's the earlier Javalobby discussion on this subject.

(Note from Rick: there's some confusion about what the actual status of the situation is, and I have asked Sun for some help to clarify. Will post updates as soon as any are available...)

JXMLPad is a pure swing component/framework for editing XML/XHTML content.
Home :
http://www.japisoft.com/xmlpad

Sketsa is a vector based drawing application based on SVG. Scalable Vector Graphics (SVG) is a graphics format and technology based on XML developed by W3C. With Sketsa, you can create vector graphics that can be scaled and printed at any resolution,

SearchBlox is a J2EE search component that enables you to add search functionality to your applications, intranets or portals in a matter of minutes. SearchBlox features integrated HTTP and File System crawlers, support for different document formats,

db4o announces release 2.8 of "db4o - database for objects".

This release is the result of a thorough internal refactoring, and it comes with a large improvement of query execution speed at a lower memory footprint.

Hi all,

JIDE Software continues expanding product line and marches towards the goal of a complete client side Java/Swing solution. Today we are happy to announce the availability of a new product - JIDE Dialogs. One of the major components in this

The Javio Swing Component Suite 1.5 is available.
The Component Suite is a library of Swing GUI components that can easily and quickly enhance the user experience of any Swing application. The components use advanced painting techniques in order give

This one seems to be snappier than the first version with the new UI. I really like it. GO NETBEANS!

http://www.netbeans.org/downloads/ide/development.html

The UIHierarchy library aims at simplifying the development of user interfaces made in AWT or Swing for Java 1.2 and later.
It simplifies code to match the mental picture of containement hierarchy, which makes it easier to develop and maintain small

The MX4J project has released MX4J 2.0 beta 1, an open source implementation of the JMX technology (JSR 3) version 1.2.1 and of the JMX Remote API (JSR 160) version 1.0.

With this release, a complete implementation of JSR 3, version 1.2.1 and of

The Struts Console is a FREE standalone Java Swing application for managing Struts-based applications. With the Struts Console you can visually edit JSP Tag Library, Struts, Tiles and Validator configuration files.

Jameleon is an automated testing tool that separates applications into features and allows those features to be tied together independently, creating test cases. These automated test cases can then be

Cortex is pleased to announce the release of Clover 1.2.4, and version 1.1.3 of the Clover plugin for Eclipse.

Clover is a Code Coverage tool for Java. It finds sections of your code that are not being tested. This helps you improve your unit

A new version of JFreeChart has been posted at:

http://www.jfree.org/jfreechart/index.html

Try it today and find out why developers are switching to JFreeChart.

Magnolia 1.0

Obinary releases free Java-based Enterprise-Content Management System (CMS) - more than 1200 downloads within first 4 days

UploadBean is a JSP/Servlet component allowing to upload files from any browser and store them into a folder, a ZIP or a Database (Oracle, DB2, MySQL, ...). Release 1.5 improves CFU parser integration, adds PostGreSQL support, fixes a few bugs and adds a

Hi,

As the result of the massive feedback i got, i uploaded
a small bugfixed version of the Player.

The bug reported occured during init of the player, and therefore i decided to fix this. On some systems an internal error was thrown

Hi,

The Web Start News Wire (aka lopica-announce) reports that Jonathan Buckland - a Java developer from the UK - has launched a new Java Web Start application catalog.

jTDS, the JDBC type 4 driver for Microsoft SQL Server is back on track. Before heading into some serious refactoring to allow further development, we have released a new version.

The 0.5.2 release adds 3 new features: SQL Server named instances,

Hi,

JNIWrapper 2.1 for Windows
======================

Since the JNIWrapper 2.0 release we have received a great number of feedback on the new functionality. Few problems have been found and many new interesting features suggested.

We are not