Six Years and Counting: Developers Weigh in on the State of the Java Market

Six Years and Counting: Developers Weigh in on the State of the Java Market

f you look at it from one perspective, the Java programming language at age six is only in first grade. On the other hand, Java’s significant strengths and large base of devoted followers mean that in many respects this versatile technology resembles more a seasoned veteran than a grade-schooler.

There’s no denying that Java has made an impact of unprecedented proportion in the development world and will continue to do so. And with this year’s JavaOnetrade show and expo just around the corner, DevX decided several months ago to set about the task of looking closely at the state of the Java market, with a special emphasis on where Java is making its biggest impact in today’s business environment. We also wanted to find out how our readers feel about Java’s accomplishments and failures to-date and what the future holds in store for Java developers and the businesses who rely on it to drive their success behind the scenes.

There are several components to our special report, “Judging Java.” There will be three pieces in this series: This week, we begin with our comprehensive analysis of Java based on recent private research, which you are about to read (click here to skip this intro). You’ll also want to read the adjunct article Server-side Subjects Drop Java’s GPA, in which Java expert and regular DevX contributor Brian Maso of DevelopMentor takes a critical look at what Java can and can’t do on the server side. Finally, please don’t miss Java in Retrospect, a complete timeline of Java’s history, which charts every event and nuance of Java’s development from 1995 to 2001. We believe this is the most extensive and comprehensive history of Java ever compiled. You’re sure to refer to it often, as both a source of reference and entertainment.

You’ll read our exclusive interview with Sun‘s Dr. James Gosling, the man who led the nascent Java specification to completion six years ago and who created the first Java Virtual Machine and compiler. We’ll also take an in-depth look at the Java job market. We wanted to find out whether there’s a “skills gap” emerging in the Java development community. Are there really enough Java developers with the advanced enterprise knowledge needed to develop high-availability, high-scalability applications that today’ business environment requires? You’ll find out whether your skills meet the expectations of today’s recruiters. And throughout the week, you’ll get daily reports from the show floor at JavaOne.

In Part III, launching July 11, you’ll read reviews of major products and technologies released at JavaOne. And you’ll get an early introduction to the newest member of the DevX family, Java2theMax. Java2theMax Editor John Zukowski is the founding Java guru of jGuru, served as‘s Java expert, and has written three books on Java. He is hard at work putting together a complete source of Java development resources and thought-leading editorial, tailored to the advanced Java programmer. You’ll read what John has in store later this summer for the DevX Java audience.

But first, read on to find out how more than 2,600 professional Java developers view the state of the Java market.

The State of the Java Market
Based on very recent and extensive independent research we’ve drawn a sharp picture of what Java looks like, six years out, using as our eyes those who know best: in-the-trenches professional application developers, project managers, IT professionals, CTOs, and independent consultants. With the help of the Adams Company of Palo Alto, Calif. we conducted a lengthy survey of Java professionals, polling readers of both DevX’s Java Update newsletter and Java Promagazine during March and April of 2001. The individuals who answered our survey comprised a wide range of software development capacities—19 percent were commercial software developers, 26 percent did either custom/contract applications or worked at a commercial ISV, 24 percent fell into the in-house corporate development group, and 13 percent worked at application service providers.

To learn more about the demographic makeup of our survey respondents, their professional experience, and job titles, please see our sidebar “The People We Polled.”

Despite the somewhat tedious length of this survey, more than 2,600 people responded, telling us about where and how they use Java, what they like and don’t like about Java and competing technologies, what they see in Java’s future, and their opinions on the various Java products and vendors in the marketplace. It encompasses tools, operating environments, projects, teams, skills, name brands, reputations, preferences, demographics, and a lot more. (Percentages in the text of this article have been rounded to the nearest whole number. For exact tallies see the charted data.)

Are Tools So Cool?
Java may have finally broken through the majority of barriers that held it back in its early years thanks to improved performance and greater fulfillment on the promise to run anywhere consistently, but if there’s one thing that’s clear from our research, it’s that tools to make Java development easier at every level of the programming hierarchy could still improve.

Our readers told us that they see a significant gap between what they need from Java tools and what they get from the Java tools currently on the market. In our survey, we asked respondents to rate the importance of certain tool features on a scale of 1 to 5, with 5 being the highest importance ranking. Figure 1 shows the number of respondents giving each feature either a 4 or 5 ranking. Figure 2shows the number of developers saying they gave a 4 or 5 ranking to the current stable of available tools for Java development.

Java developers want simple features that work. Performance ranked highest among 16 different features that developers usually rely on commercial tools to provide. Eighty-six percent answered that performance is of critical importance in a Java tool, followed by conformance to standards at 84 percent and debugging at 83 percent. The next-most named features that our respondents said they longed for were code reuse and speed (both 79 percent), followed by a graphical user interface (71 percent) and XML support (69 percent).

Figure 1. Importance of Certain Tool Features: Percentage of those saying that the feature listed is a 4 or a 5 in importance (on a scale of 1 to 5)
Figure 2. Satisfaction with Tools: Percentage of those saying that their satisfaction with the feature listed is a 4 or a 5 (on a scale of 1 to 5)

Judging by Figure 1, rapid application development features, such as automated coding, are not valued particularly high by today’s Java developers; only 32 percent gave it a 4 or 5 rating. Further, developers don’t seem particularly anxious to find a one-stop IDE for development. Only 37 percent felt that it was important for coding tools to be sold as a suite (see also Figures 4 and 5).

What we noticed immediately in looking at this data was the significant drop in those respondents giving a 4 or 5 rating to satisfaction with the available tools. Compilers, at 55 percent, are the only tools that the majority of developers felt are currently exceeding their expectations. SDKs received the next-highest ranking at 48 percent. IDEs were third (36 percent). Database tools and application servers were a precipitous step lower.

Modeling tools, deployment tools, team development tools, and report generation tools brought up the rear. We attribute low rankings across the board to general immaturity in the Java tools market.

Figure 3. Importance of XML Support: Importance of XML support in Java tools as an influence on business development over the next 12 to 18 months, broken out by: a) all respondents b) those who program primarily in Java and c) those who said they were highly skilled in XML

One of the biggest questions this brought to our minds is: Are top-notch Java development tools rare because Java developers don’t demand them, or have Java developers learned to live just fine, thank you, without top-notch toolsets that other language developers have come to expect because the tools were too slow to materialize? The answer is probably a little of both. Let’s face it: Java programmers, in general, are a sophisticated lot, most of whom have traditional computer science training and deep knowledge of general application logic. You’ll see more evidence of this throughout the results.

XML support was among the things consistently ranked very high in importance (see Figure 3). Not surprisingly, those respondents who gave themselves high rankings in XML expertise deemed it most important with 41 percent giving XML support a 5 ranking. Those who said Java was their primary programming language came in a bit higher than the status quo, at 36 percent vs. 34 percent for the entire respondent base. Indeed, our study showed a strong correlation right now between Java and XML, with a strong increasing trend over two years.

We wanted to know how Java developers were using specific tools to develop and deploy applications. And we wanted to know how they felt about the current product offerings within each of the major tool categories. To do this, we asked respondents a series of questions designed to find out:

  • Comparative market reputation of each product
  • Comparative customer reputation of each product
  • Market share of each product
  • Expected net change in market share over the next year

Of the questions in this series, integrated development environments (IDEs) yielded the most interesting findings. Only 65 percent of the total respondent base uses any IDE at all. A fair number of Java developers, it seems, would rather hand-code in Emacs or possibly Notepad and then use a standalone tool for compiling. Debugging is often performed by printing variable contents to the screen. Of those developers who do use an IDE, few restrict themselves to only one.

Figure 4. Reputation of IDE Tools: Percentage of those who gave each product a 5 rating (on a scale of 1 to 5), broken out by a) all respondents and b) respondents who use the tool
Figure 5. IDE Current/Future Use: This pink area shows the current marketshare for each vendor/product listed, by percentage. The green area shows the net change in marketshare over 1 year, based on the number of respondents who said they would adopt, stay with, or abandon that product line.

Borland is still the name to beat. Sometimes we wonder whether any vendor will ever eclipse Borland’s long and amicable friendship with the developer rank-and-file. Nearly 20 percent of our respondents gave JBuilderthe highest ranking of 5. A whopping 34 percent of respondents gave the product a 4 (data not shown).

Allaire’s JRun Studio and Oracle’s JDeveloper ran neck-and-neck in our survey. Surprisingly, JRun edged out JDeveloper in many metrics; though the difference was minuscule, most developers said they had a more favorable view of JRun than of JDeveloper. However, JDeveloper did edge out the Allaire (now Macromedia) product slightly in current users, with 17 percent to 15 percent, respectively. The JRun application server also made a healthy showing in our survey (see below).

Tomcat Mania
Java developers are embracing the Apache-based Tomcatapplication server in a major way, and why not? It’s free. The highest concentration of Tomcat users are those who said they primarily build client-server applications (49 percent use Tomcat, data not shown), multi-tier applications (48 percent), and corporate enterprise applications (also 48 percent), though usage was in the high 40s across the board, regardless of concentration.

Tomcat is no-frills, but our research leads us to believe that may be exactly the way developers like it.

The Sun/Netscape Alliance’s iPlanet app server made a mediocre showing in our survey, both in reputation (only 5 percent ranking it a 5—see Figure 6) and future adoption plans (3 percent said they would move to iPlanet in the next 12 months—see Figure 7).

Perhaps the biggest news of all is that the JRun application server is close behind both BEA’s WebLogic and IBM’s WebSphere. JRun hasn’t had a lot of time to gain maturity because Allaire only acquired the technology as part of its purchase of LiveSoftware in 1999, but the product’s low entry price has surely been a key factor in putting this product squarely on the map quickly. It’s also been well reported as a product that’s notably easy to install and use. However, note in Figure 5the low number of respondents who said they would be adopting the product in the next 12 months. Net market increases for BEA and IBM look far healthier.

Figure 6. Reputation of Application Servers: Percentage of those who gave each product a 5 rating (on a scale of 1 to 5), broken out by a) all respondents and b) respondents who use the tool
Figure 7. Application Servers Current/Future Use: This yellow area shows the current marketshare for each vendor/product listed, by percentage. The blue area shows the net change in marketshare over 1 year, based on the number of respondents who said they would adopt, stay with, or abandon that product line.

Next we asked about messaging and transaction processing middleware. No surprise here. IBM’s MQSeries leads the field, with BEA’s Tuxedo second. Far afield is Sun Microsystems with the relatively new JMS system, which is an also-ran with Progress (Sonic Software division) and Fiorano.

Interesting to note, though, that actual users of MQSeries generally rated the reputation of the product higher, compared to the full response group, than other products. In other words, while MQSeries in general received a highly positive 5 rating from 17 percent of respondents, a full 20 percent of those who use it gave it a 5 rating.

Tuxedo users were only 1 percent off the mark for reputation in general, while Sun’s JMS faired less well among its users than it did with the full respondent list, perhaps indicating that JMS needs some time to mature, though it had better do so fast: Figure 9shows that JMS will double its market share in 12 months.

Figure 8. Reputation of Messaging Tools: Percentage of those who gave each product a 5 rating (on a scale of 1 to 5), broken out by A) all respondents and B) respondents who use the tool
Figure 9. Messaging Tools Current/Future Use: This green area shows the current marketshare for each vendor/product listed, by percentage. The pink area shows the net change in marketshare over 1 year, based on the number of respondents who said they would adopt, stay with, or abandon that product line.

Model of Success
In Java modeling tools, Rational Rose, long the de facto ruler of the modeling toolkits, has, not surprisingly, run away with the field. The next-most respected product in our survey, TogetherJfrom TogetherSoft, will have trouble closing the gap. Fifteen percent of all respondents said they would add Rational Rose to their software lineups in the next 12 months while only 9 percent plan to adopt TogetherSoft.

The third most recognized product in the field was Sybase PowerDesigner, followed by Embarcadero Technologies’ (formerly Advanced Software Technologies) GDPro.

Figure 10. Reputation of Modeling Tools: Percentage of those who gave each product a 5 rating (on a scale of 1 to 5), broken out by a) all respondents and b) respondents who use the tool
Figure 11. Modeling Tools Current/Future Use: This green area shows the current marketshare for each vendor/product listed, by percentage. The blue area shows the net change in marketshare over 1 year, based on the number of respondents who said they would adopt, stay with, or abandon that product line.

On average, our respondents said that they built 43 percent of their applications in the Java language (see Figure 12). After Java, Visual Basic was the language most often cited for application development in our study. XML (though not a language in itself, we felt it useful for comparison in this context) and C++ each received 22 percent.

Both VB and C++ development will fall off in usage during the next two years, while XML development will increase more than any other language named. After two years, our respondents estimated that their use of Java will increase by 5.7% to 49% and pure XML development will increase by 6.2% to 28%. In Figure 14, we looked at the use of XML specifically in conjunction with Java and saw a far more precipitous increase in usage predictions.

Sun Microsystems Java evangelist Simon Phipps said it rather well at a recent conference: “You can have all the XML you want flying through the air from place to place, and that’s fine and good, but there needs to be Java on a machine at both ends to make it actually work.” It appears that this message of Java and XML’s highly interdependent relationship is well understood by developers.

Wireless application development will increase by about 70 percent over the next two years, an indication that J2ME may be a more sought-after platform in the months to come.

Figure 12. Apps Involving Language/Technology: Percentage of total applications developed in each language, broken out by A) current development and B) expected development in 2 years

Which Version of Java Prevails?
It’s clear from Figure 13 that most developers don’t use any single version of Java exclusively. After J2EE and J2SE, the next-highest number of users, 33 percent, was for the JDK 1.1.1. We presume that many developers in this category are IBM shops that haven’t yet upgraded to version 3.5 of WebSphere, the first J2EE-compliant version of IBM’s Java environment. But another logical explanation is that there is a prevailing desire to ensure applications work with the largest common denominators, which means paying special attention to Microsoft Internet Explorer.

Eighteen percent of Java developers use Microsoft’s Visual J++. Many Windows developers turn to Visual J++ when tight integration with COM objects is needed. This is where Visual J++ excels, and because it is provided free with the Visual Studio suite, we’re not surprised to find some people using it where it makes sense.

Figure 13. Versions of Java in Use: Number of developers using each of 6 different Java versions, by percentage (multiple responses allowed)

We asked developers to tell us what technologies they used most often in developing their Java apps. JSP/servlets came in highest at 40 percent, with XML next at 30 percent. Developers said within two years they expected XML to overtake any other technology as a Java facilitator, increasing 21 percent in their use of XML over that period. If their predictions hold true, 51 percent of their applications will involve some use of XML for moving data from one location to another. Java-based WAP development is expected to increase from 6 percent currently to 17 percent in two years.

Figure 14. Technologies Used with Java Apps: Percentage of Java applications using certain enabling technologies, broken out by A) current development and B) expected development in two years, by percentage

We figured it would be a nice thing to ask our respondents what it is their Java applications are designed to do. The answer, of course, was just about everything. A whopping 14 of 16 categories garnered double-figure percentages, including Web-based data entry at the high end (63 percent) to ERP apps (12 percent) at the low end (see Figure 15).

What don’t you use Java for? The traditional answer to this question is: “You don’t develop things like drivers in Java,” although Java creator Dr. James Gosling believes otherwise (he discusses this in our exclusive Q&A, coming later in this series).

In particular, database access and management applications (52 percent) were cited most often as the impetus behind non-Java language development (see Figure 16). Database development is simply not as convenient to do in Java as it is in other languages; performance issues and compatibility with non-Java legacy data are factors that play here as well.

Other non-Java apps that were named in Figure 16: Office applications (36 percent), transaction processing (35 percent), and Web-based data entry and retrieval (34 percent). It’s noteworthy that 14 percent said they were using languages other than Java for embedded systems development, as opposed to only 9 percent who said they were using Java for embedded systems.

Figure 15. Principal Function of Java Apps: Percentage of Java applications developed for the business function described (multiple responses allowed)
Figure 16. Principal Function of Non-Java Apps: Percentage of non-Java applications developed for the business function described (multiple responses allowed)

One of the goals of our special report has been to find out whether the skills and talents imbued in the average Java developer are sufficient to meet the demands of today’s enterprise business needs. We will debate in detail the possibility of a widening Java skills gap in the second installment of our “Judging Java” special report, to be published the first week of June.

But in our survey, we asked respondents to rank their current skills in each of nine categories on a scale of 1 to 5. Figure 17charts just the 5 responses. Database development was highest, with 17 percent of respondents giving themselves the high score. Java, appropriately, was next with 15 percent saying they were a “5.”

We then asked respondents to tell us how important they thought each of these skills would be to employers during the next two years (Figure 18). Nearly half (48 percent) of respondents gave Java a ranking of 5 as an important skill category to employers during the next two years, while 44 percent gave XML the 5 rating. The next-highest ranking was database applications at 31 percent. Most other skills listed received far fewer “5s.”

Figure 17. Level of Expertise: Percent of developers who gave themselves a 5 rating (on a scale of 1 to 5) as to their expertise with the language or technology listed
Figure 18. Demand for Skills: Percent of developers who gave a 5 rating (on a scale of 1 to 5) to the importance of each language or technology listed in the development job market
Figure 19. The Skills Gap: The black numbers represent the value that developers placed on this skill in two years’ time. The green area represents the current expertise to meet this demand in the marketplace. Blue shows the gap in supply and demand.

By comparing Figures 17 and 18, we tried to determine just how well-equipped developers are to handle the current demands in the marketplace—based on their own opinions. The results (Figure 19) show that indeed there may be a real skills gap when it comes to the highest level of expertise in key categories. XML development showed a more than 40 percent gap between currently skilled developers and upcoming hiring demand. Also high was Java itself, with a 33 percent potential gap in available expertise.

Wrapping It Up
Six years and several million applications later, Java is proving itself a world-weary veteran despite its tender age. Of course, there’s a lot more to “Judging Java” than can be revealed through research. Read on to the second piece in this series, Server-side Subjects Drop Java’s GPA, for a very thorough, technical examination of Java’s significant abilities and disabilities in server-side application processing. Author Brian Maso gives a perfect, encapsulated analysis of what today’s businesses and technology implementers need to get from Java next.

For further reading relevant to a thorough analysis of Java, please see the Other Resources links.

Written by DevX Editor in Chief Lori Piquet, with assistance from DevX Senior Editor Chris Preimesberger and Java Pro editors Sean Gallagher and Steve Gillmor.


About Our Editorial Process

At DevX, we’re dedicated to tech entrepreneurship. Our team closely follows industry shifts, new products, AI breakthroughs, technology trends, and funding announcements. Articles undergo thorough editing to ensure accuracy and clarity, reflecting DevX’s style and supporting entrepreneurs in the tech sphere.

See our full editorial policy.

About Our Journalist