Gartner says companies overspent $1 billion on app. servers

Discussions

News: Gartner says companies overspent $1 billion on app. servers

  1. The Gartner group has released a report indicating that companies have overspent about $1 billion on application server technology solutions since 1998. Moreover, an additional $2 billion may be wasted between now and 2003. Gartner estimates that, by 2003, 60 percent of all new J2EE application code will remain JSP/servlet-only.

    Press Release:
    -----------------------
    STAMFORD, Conn.--(BUSINESS WIRE)--Aug. 21, 2001--Gartner, Inc. (NYSE: IT - news and ITB - news) has released a report indicating that companies have overspent about $1 billion on application server technology solutions since 1998. Moreover, an additional $2 billion may be wasted between now and 2003.

    Gartner makes the following recommendations for companies when they purchase and implement application server technology:

    Application server is a function, not a product. Check your technology inventory to see if you already have the capability in an existing product or a free product.
    Application server is not the same as Java 2 Enterprise Edition (J2EE). For example, the Microsoft Product line lacks J2EE and Java, but has application server functionality.

    As components of application server technology, J2EE and Enterprise JavaBeans (EJB) are not the same thing. Most Java projects use Java Server Pages (JSP)/servlet capabilities and not EJB. Higher-priced application servers are designed to run EJB, yet they are using JSP/servlet capabilities instead.

    Don't let confusion or hype push you to spend more than necessary.

    Gartner estimates that, by 2003, at least 70 percent of new applications will be deployed using high-end application servers and 60 percent of all new J2EE application code will remain JSP/servlet-only.

    ''The application server vendors are encouraging customers to purchase higher-end technology that they just don't need,'' said Gartner vice president and research director David Smith. ''It's like buying gourmet food to feed kids at summer camp. It's just not necessary.''

    About Gartner, Inc.

    Gartner, Inc. is a research and advisory firm that helps more than 10,000 clients understand technology and drive business growth. Gartner's divisions are Gartner Research, Gartner Consulting, Gartner Measurement and Gartner Events. Founded in 1979, Gartner, Inc. is headquartered in Stamford, Connecticut and consists of 4,600 associates, including 1,400 research analysts and consultants, in more than 80 locations worldwide. The company achieved fiscal 2000 revenues of $855 million. For more information, visit www.gartner.com.

    Threaded Messages (54)

  2. Sad, but true. I'm thankful that the project I'm on chose WebLogic for the good job experience, but they don't even closely need an app server and an oracle database; the largest table is less than 200000 rows and they don't even us transactions, the database is largely read-only. They were sold a bill of good is what happened and spent $1Mill when they could have spent around $100000 (counting development time).
  3. After being on several consulting engagements I have found that many project managers, CIO, and Vice Pres. are not in touch with the J2EE community. There are several cases where an application server is a must.

    [ Comment 1 ]
    **********************************************************
    Application server is a function, not a product. Check your technology inventory to see if you already have the capability in an existing product or a free product.
    Application server is not the same as Java 2 Enterprise Edition (J2EE). For example, the Microsoft Product line lacks J2EE and Java, but has application server functionality.
    **********************************************************

    [Rebuttal]
    An application Server is NOT a function it is a product. I think what you are trying to say are many application servers having similar functions can lead one to believe there is no difference when choosing one.

    In a high traffic production system the Application server serves as the base infrastructure for all web related functionality. With out this infrastructure many of the services listed below would need to be written from scratch or purchased in individual pieces.
    * SERVLETS ENGINE
    * JSP's SUPPORT
    * SERVLET SUPPORT
    * LOGGING
    * DB CONNECTION POOLING
    * EJB's SUPPORT
    * RMI SUPPORT
    * CORBA SUPPORT
    * SECURITY SUPPORT
    * CLUSTERING
    * SESSION FALL-OVER
    * THE BASE FOR ALL WEB APPLICATIONS

    If you are deploying a mission critical application why use products that are cheap and mostly unproven in PRODUCTION. I believe that over estimating and not fully understanding J2EE has cost companies like Gartner and many others millions of dollars.



  4. This is unbelievable. Is Gartner suggesting to run your enterprise production apps on free products. i agree with doug, its gartner that needs to understand the subject of J2EE and app servers comprehensively before making such a statement.

    i agree companies may have overspent on certian things, may be even app servers. anyway just tell me one thing, why cant u run a pure JSP/Servlet enterprise app on a App server ?

    i think Gartner should stick to stats n figures, rather than technical evaluations, if they dont know what they are talking about.
  5. I do not know where they come up with the number. To my knowledge, the cost of J2EE application server is relatively cheap compared to what a project really costs you. A decent project (in terms of size and importance) easily costs more than a quarter of million.
  6. I do not know where they come up with the number. To my knowledge, the cost of J2EE application server is relatively cheap compared to what a project really costs you. A decent project (in terms of size and importance) easily costs more than a quarter of million.
  7. I think you miss the point - no one would argue that J2EE, EJBs and a high-end application server are appropriate for very complex sites (although I might argue that J2EE and EJBs have not been shown to be the highest performing platforms with the exception of Orion).

    But the number of "high-traffic sites" these days is pretty darn small, and, in most cases, EJB is overkill for the functionality to be created.

    I think what Gartner is really saying is this: Microsoft is shipping an application server (IIS + ASP + COM) that will satisfy most of your site's needs and you've already made the investment because of those Windows 2000 servers sitting in the rack already. I believe it was Gartner who also estimated that Microsoft had the ability to control 93-95% of a company's computer environment - leaving the high-end sites to Sun and BEA. So, there may be a hidden agenda here.

    When I have a new client, they've often come in with a headstrong IT person who completely believes in the J2EE App Server to Oracle setup. I caution them to look around - from free solutions (Orion, Enhydra, PHP) to sort of free solutions like Microsoft's ASP (free with Windows 2000), to lower-end Java solutions like ServletExec - there are a lot of choices that may save the client a lot of money and provide the same functionality.

    I just think that J2EE application servers are way overpriced - $20k per CPU is way too much. But the market will correct itself as IT spending continues to slow down and new players enter the market.
  8. let's look at all the "value" an application server gives:

    * SERVLETS ENGINE
    * JSP's SUPPORT
    * SERVLET SUPPORT

    these three are the same thing. take an httprequest and write strings to the page. use tomcat. cost: free

    * LOGGING

    what app server has a good logging package? use log4j. cost: free

    * DB CONNECTION POOLING

    i'll email you one. it's a stack of connections that we automatically close when placed back in the pool. cost free.

    * EJB's SUPPORT

    use jboss if you think ejb development/deployment is really something your company wants to fight with. if you use BMP you really aren't using much ejb funcationality anyway. cost: free

    * RMI SUPPORT

    serializing objects and class definitions? i won't even comment on this one. rmi is for suckers.

    * CORBA SUPPORT

    corba support? that works? have you heard of webservices?

    * SECURITY SUPPORT

    has anyone out there used the ejb security model? i doubt it.

    * CLUSTERING

    does the data syncronization work? is it more overhead than it's worth? should you need to cluster for a 500 concurrent user site?

    * SESSION FALL-OVER

    99.99% of all ejb's are stateless sessions.

    total cost of a production system: free.

  9. Companies overspent $1 billion on app. servers ?

    Doesnt sound much to me...

    I wonder how much they overspent on:
    1) Ill-conceived projects (err.. 90% of all .coms)
    2) Results of poor technical management
    3) Results of poor quality control
    4) Re-inventing the wheel (read:framework)
    5) Supporting 100's of low-end servers instead of fewer big-iron servers
    6) Expensive consultants
    7) Funky office furniture

    There are some extreme stats that say 90% of all IT projects fail (to meet deadlines, budget, requirements). Given that statistic - the overspend on appservers is chicken feed.

  10. Colin,

    1.
    * SERVLETS ENGINE
    * JSP's SUPPORT
    * SERVLET SUPPORT

    Did you try to compare the performance of the "free" servlet engine and commercial servlet engines like Weblogic?

    Did you ever try to get support from your "free" Servlet engine? Probably you would get a answer "use the source, luke.".

    2.
    * DB CONNECTION POOLING
    Did you ever look at complete properties of connection pools in advanced containers like Weblogic? Are you ready to mail me a connection pool code which has all the properties and provide support?

    3.
    * EJB's SUPPORT

    I don't have much expereince with free EJB containers but This is not one I would take a cost free one and mess up my system.

    4.
    * CORBA SUPPORT
    * RMI SUPPORT
    I suggest you to look at specs and real world application which use RMI and CORBA.

    Thanks,
    Kumar.

  11. 1. let's see, you can PAY for iPlanet and what servlet/jsp motor will you get? tomcat. if i were weblogic, etc. i would have published data showing how much faster my webserver was than the tomcat/servletexec/jrun etc but they don't. do you know weblogic's is faster? if not, the point is moot.

    what support do you need? every time i've tried to use the pageContext or get a value of the request, it's worked. one other way to go is jboss and if you need support, you can pay one of their consultants, but he appserver/bundled tomcat are free.

    2. it's a connection pool. keep it simple. what properties do you need? i can tell you, max and min. that's it. there's nothing more a connection pool can do for you.

    3. why do you assume ejb's are hard? when 99% of apps are stateless session/BMP what the hell can go wrong? the executor thread dies?

    4. i understand the specs. one of the points of the article was people aren't using anything other than a vm to construct webpages. what's the future of computing as of today? p2p computing using xml (with or without soap). why in the heck would i be pushing class templates across the wire?

    i like to develop and make systems work. it's not about bulletpoints and what if's and maybes. it's about solving business/technology problems.

    if you want to pay $50,000 for a vm, have at it. i think there's better ways for any development team to spend the company's money (like training).
  12. This report smacks of Microsoft disinformation. For a long time now it's been Microsoft's line that app servers are not needed for any project. Why? Because they don't sell them. App servers are almost exclusively Java based and we know Microsoft's attitude towards Java.

    A major reason why JSP/Servlets are still dominant in many projects is it's a known element. The environment is stable and many developers have been using them for some time. It's also true that from the backend side there are many developers who are capable of developing database systems using JDBC. But this is changing and EJBs are part of the mix of distributed services that make building complex systems less likely to fail. I think BEA and IBM are showing that integrated services with a large number of choices of how to implement them is the way to go.

    The bottom line is not really whether you have to use EJBs or not it simply depends on the project.
  13. Windows 2000 / COM+ _is_ the app server that Microsoft sell. And as mentioned elsewhere, using the built back-office functions such as IIS, MSMQ and XML parser can be a very cost-effective option to the business.

  14. For a longtime now it's been Microsoft's line that app

    >servers are not needed for any project. Why? Because they
    >don't sell them. App servers are almost exclusively Java
    >based and we know Microsoft's attitude towards Java.

    It is correct that MS doesn't sell an app server product but it is not correct that MS doesn't offer any app server functionality. App server functionaly comes with W2K, but it is MS proprietary and not at all J2EE compliant. It also not correct that MS argues that app servers are not needed in projects. The MS architecture (just like J2EE) is a multi tier architecture and in this architecture the business logic is positioned on a app server.... In my opiniun app servers are not exclusively Java based, J2EE app servers are exclusively Java based but that is obvious. Even before J2EE there where app servers: eg Corba based application servers, Tuxedo based application servers, some people even consider our mainframe with CICS transactions as an appliction server......... One can state that J2EE based app servers are better than mentioned alternatives but that is different discussion......
  15. Colin,
    1. For sure I know Web logic has one of the Best servlet containers.
    I Don't work for BEA. So I don't think I am authorized to publish any comparison results.
    Regarding support propabably 95% of the time you use basic stuff that works with every container but oneday for sure you are going to make a support call.
    2. Would you mind looking at weblogic connection pool documentation?
    I would like to have properties like:
    a) Release my connection during off peak hours so that some other batch process can use connection.
    b) When my DB goes down and up again automatically detect it and create fresh pool.
    .......

    3) Developing EJB's is not hard. But Imagine you the situation you EJB container doesn't handle transactions properly.

    4) IF you want to pay waste a million for wasted development effort just because you made a poor choice of containers you free to go ahead.

    Kumar.
  16. 1. if you know weblogic has a "faster" servlet engine tell me. seriously. i'm not looking for whats and maybes. if it's faster (and tomcat isn't fast enough) let me know please.

    2. i've used the weblogic connection pool. in two ways your missing the point: a) i'm not speaking of weblogic vs. jboss. i'm talking about all the crappy implementations (that cost money) vs several very viable free containers. the properties you're talking about aren't part of the j2ee spec. if you want to use them, more power to you but the promise of portablity is lost. b) what are "off peak hours" and why wouldn't you use another pool (or no pool) to do batch processes? connect to the db and run the batch process. keep it simple. as for the up/down db, if it *works* that certainly is a helpful thing.

    -- as an aside --
    db connection pools aren't part of the j2ee spec at all. how you attach/pool (if at all) any resource is up to the container provider.

    3. there's nothing intrinsic better about the transaction management code that you get as your price paid increases.
    the reason why you unit test your code in the container is to assure the application works (in the container).

    4. the value of j2ee containers is, if there's an issue with the container (and don't use services that are outside of the spec) you swap it for another one and run your unit tests. there's no value in paying for the code to run your code.

    there's value in making sure your application works. my application works in jboss/tomcat.
  17. Colin,
    Tomcat is well known to be very slow. Perhaps all the industry participants are reluctant to release their best code in this reference implementation.

    Orion published an article on the slowness of Orion. Meanwhile, new comers like Rexip are claiming even better performance than Orion and JRun.
  18. I agree!!! Support and good documentation is everything. Products like weblogic have great support and even better documentation. Gartner just pointed out or picked on one service of J2EE, and that's EJBs. Maybe their prediction will come true, but EJBs are not what you spend all your money on. I wish they would evaluated all J2EE services and other core services that you are paying for instead of picking on one service.

    Everyone lately feels the need to put down EJBs and that's fine, because sometimes using EJBs is like crushing a walnut with a sledge hammer. Whether using EJBs or not, in my opinion, shouldn't be the only reason to stay away from a commercial application server.
  19. If RMI and Corba are not effective approaches, what are your recommended alternatives?
  20. my issues with rmi:
    in general:
    95% of the time all you care about is moving data from one vm to another. why do the class templates have to be the same? anyone out there have IncompatibleClassChange issues?
    in ejb world:
    99% of the time calls between ejbs are local, so why always serialize? i've heard the new ejb spec addresses this issue but i think it's up the the deployer to explain what calls are intravm. in my current project, zero of our classes implement serializable.
    now, i've heard veeners say the when calling a method via rmi you're using a contract (with well defined semantics) where using a webservice only gets you "data". i'm still on the fence as to whether these are different.

    what's better?
    www.glue.com
    seamless intra-vm services with soap-wrappered xml via port 80 intervm communication. and if you continue to use a service over the wire? glue will move it onto a local node. voyager was as cool and now this uses soap/xml on http as the protocol. you may have to define your services with a larger granularity (or facade some current services) but seemless xml proxying seems better than requiring rmi registries, ports, and rigid class definitions.

    corba:
    i've been developing for over four years and have yet to see an example of a good use for corba. when i've had to talk with a desperate system, i talk xml. in all cases, there's always been a component boundry and the cleanest way to decouple them was not to use a type-type mapping but to trasfer and use self-describing data...in one case we got reuse out of the component for a delphi developer.

    cheers,

    csb
  21. Colin,

     FYI: Weblogic allows you to use pass by reference and hence avoiding serialization.
     

     Kumar.

  22. Oh My...they just had to come up with something...rebellious right.. if not anything else...If Giga is bullish on the J2EE space then we have Gartner who pulls it down...

    In my opinion apart from Product vendors; firms which make money by selling out reports on technology and more often mislead the market than anything else are absolute farce.

    Another junk of a report...like the million others that you can find in the market.
  23. Another junk report..[ Go to top ]

    Oh My...they just had to come up with something...rebellious right.. if not anything else...If Giga is bullish on the J2EE space then we have Gartner who pulls it down...

    In my opinion apart from Product vendors; firms which make money by selling out reports on technology and more often mislead the market than anything else are absolute farce.

    Another junk of a report...like the million others that you can find in the market.
  24. again, i am speaking to the SPECIFICATION. weblogic can through anything they want but using it ties you to one app server vendor. not good.

    do you have to use t3 or rmi? do you have to, at deploy time, tell the app server where the services will reside?

    thanks,

    csb
  25. Colin's easy answers:

    Phones:

    Who needs them? Just go next door and talk to your neighbour.(FREE) If anybody is far away, write a letter.

    Printers:

    Use a pencil. 5 cents.

    Computers:

    What a waste of ressources! Everything you can do with a computer can be done with a brain also. You got one, use it.(FREE)

    Cars:

    Walk there, feet are free! If going to a far away place, take a train, it's usually a thousand times cheaper then buying a car.


    Now Colin, how about a change in your argumentation. You could become amish, so pc would forbid me to tease you....

    Regards,
    Benny
  26. benny,

    i'm not sure how you got so far off track. i'm not saying "don't use technology". i'm saying over 90% of applications out there don't use CMP, stateful sessions, nested transations (or anything other than Connection's setAutoCommit method), the ejb security model, etc etc., so why spend money on the support for these things? that's all.
    like i said before, you can PAY for iPlanet and you'll get tomcat's servlet motor running your servlets (including jsps). where is there value in that? the vm is free!

    i have one question: if ejb developers were paid 30% less than normal java developers, would you give a crap about ejbs?

    csb

  27. Very true - I got some numbers like 75% J2EE deployments
    which are JSP/Servlet apps and EJB takes the rest 25%.

    And many companies spent average 700K for BEA's WebLogic
    and IBM's WebSphere to do just JSP/Servlet apps. To do
    a simple math, I can see $1 billion USD is underestimated
    by Gartner.

    Tieu Chu
  28. I'm not doubting the Gartner group's figures however I agree with the sentiment expressed above that you really do get what you pay for. If IT managers have bought poorly then they need to look to themselves. IMHO as the EJB standards settle down, tools mature and people become aware of the features which set it apart from JSP (and other non J2EE solutions) it will take off.

    I'd love to hear what the gurus at Gartner would say about Oracle and DB2 and the % of their features which are used by developers...
  29. Don't pay much attention what Gartner is saying or trying to let you think. If you think app server is appropriate for your project use it other wise not, it is very simple. At the end of the day you are the one who has to make the decision not Gartner. Just keep in mind these reports are more political than factual. Of course app servers are not meant to be a silver bullet like any other technology.
  30. It sounds 100% true. Does all the Enterprise applications need "EJBs" ???
    If the answer is "no" then why appservers like "weblogic", "websphere" are in the top rung commanding close to 65% of the market share.So it means,for any J2EE project ,people are going in for these "Top Notch" servers irrespective whether they want it or not.
    One reason why these "Clients" are going for these might be that they are thinking that in future they might need it.In addition to this ,the marketing hype that has forced these "Clients" to go in for these "Premier" ones..without knowing whether they can use it fully or not...Hope atleast in the forthcoming years.. things might improve....
  31. I agree with gartner that for 'some' projects you do not need the complete enterprise functionalities.
    But what if you are planning a enterprise-wide (company-wide) software architecture ?
    I don't think in that case you want to jeopardize a project by using free poorly supported products (some commercial ones also have poor support :-( ). What if you need EJB in the future you need to change products and that will cost a lot.
    If a company has a vision on its software architecture they need to invest in proven technologies with good support and products wich support the whole range of enterprise features to insure future development and possible extensibility.
    I don't understand why people and companies still have a short term-vision on applications. It seems to me that everything we nowadays develop are temporary solutions with no possibility for reuse ...
    Kris
  32. Well let's face it. If you're an IT manager with your job on the line based upon the success or failure of your J2EE project, would you go with someone like JBoss and save the company 20k or would you go with WebLogic and at least have the ability to take advantage of their tech support :) (nonexistent though it may be). People don't want to see you ave 10 or 20k here and there. Most distributed applications cost in the many hundreds of thousands of dollars to develop and what difference does it make if you save 20k here or there. Managers would much rather spend a little money and be able to lean on the big names of IBM or BEA and blame them for the failure of their projects.
  33. why don't spend this money in thinking a bit more about the project ...

    many project fail because of the wrong choice of technologie , at least whitout consulting there developpement team ..


  34. >>>>>>>>>>>>>>>>>>>>>>>>>><<
    Managers would much rather spend a little money and be able to lean on the big names of IBM or BEA
    >>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<

    How about leaning on the name of SAP *and* saving those large prices? (780?/processor)
    Come to http://www.inqmy.com ...

    Sorry for the advertisement, but I NEEDED to say the truth (And Yes, I work for that company).
  35. I know my message is off the subject, but does anyone know what categories of applications are used in the financial services industry that are based on n-tier architecture?



     
  36. Well, I've worked on projects in:

    - retail banking
    - trading (matching buy/sell orders)
    - position keeping (commonly known as trading)
    - fund management

    but I'm sure there are lots of other categories where n-tier has been used.
  37. Not just BEA vs. home-grown![ Go to top ]

    Gartner is very explicit about the waste they're documenting -- companies purchase high-end versions of app servers but don't really use or need the features they provide. If you're just using JSP, sure, go ahead and buy a commercial JSP engine if you want to (JRun, Caucho, Websphere standard, whatever). But, for God's sake, don't put Weblogic Enterprise on that server!! In a lot of cases, this is really an "enterprise edition vs. standard edition" issue. Many of us have seen the same thing played out in the database world, where basic web sites get suckered into buying Oracle Enterprise, when Oracle Standard (or probably something much similar) would've been fine.
  38. Not just BEA vs. home-grown![ Go to top ]

    This is a TINY number in the scheme of things. Do you know how much companies spend overall? It's in the hundreds of billions. In other words this is <1%. They just use the absolute value because it sounds big. It's really not.

    It's also factually correct to say that they have overspent. A great many projects get done using EJB that just flat shouldn't be. After a while someone realizes and the EJB goes away (simply because it simplifies life.)

    EJB was the mount everest of technology for a while. It was used because it is there. That was it. It was back in the hight of the "technology rules" period a year or two back when anyone who could spell JAVA was paid a fortune to tout the brave new world of technology.

    Result. Well, dot.coms have gone bust, and those who can only spell JAVA now find it somewhat more difficult to find a job to support the lifestyle they are now accustomed to! :-)

    That's not MS disinformation, it's just life.

    Everyone, always overspends on new things, particularly in the finance sector where for some reason it's deemed cool to be not just leading edge, but bleeding edge!

    The fact that everyone has now calmed down and has started actually working out if the project really warrants hundreds of thousands on an app server is a good thing.

    Look at it this way. If WebLogic costs $100,000 then ask yourself this. Is the problem I am solving by buying this server worth $100,000. If so, then buy it. If not, then don't. That's an oversimplification but I don't want to get into discounted cash flows and the like here. :-)

    Chz

    Tony
  39. $1B is a big number[ Go to top ]

    TINY NUMBER?!!? You sound like a US Senator!
    No matter how you slice it, $1B is not a tiny number. Figure $1.6B for the 2000 app server market. Assume an additional $1.6B this year, so far. And add another $1.6B for 1998 and 1999 combined, giving a very generous estimate of $4.8B overall market size for the entire period in question. $1B is > 20% of that total.

    $1B is a huuuuuuge number. And the other threads have it right - $1B is probably understating it.

  40. Not just BEA vs. home-grown![ Go to top ]

    I'm with you on this.

    The point of the article is simple (and age old one), "don't use technology for the sake of using technology". You're going to waste money if you do - as they point out (to the tune of $1 Billion!).

    As far as I can see, they're not against any technology. Instead they're telling you to think about the cost/benefit of your choice (or the ROI if you wish). It's the old quality issue of defining your requirements to meet the needs of the project. For example, why do you need a to invest in a $20K server for a product that is used by 10 users periodically? Would Jboss be suitable for this (exagerated case, I know but I think that you get the point)? This is the point that others have also made on this thread

    Unfortunately, Gartner believe that we're all going to follow the same mistake again and again over the next few years - buying expensive servers for apps that don't need servers. As the aricle points out, "Higher-priced application servers are designed to run EJB, yet they are using JSP/servlet capabilities instead". In my opinion that's the real issue that needs to addressed in this thread: How can we educate people to make the right choice about servers (or as Gartner call them, 'functions') to meet the requirements of their project. What are the best practices to achieve this? By focusing on this and acting on it, we can begin to prove Gartner wrong.

    Andy

  41. Not just BEA vs. home-grown![ Go to top ]

    Unfortunately there are a lot of IT managers who, as someone on this thread pointed out, will always go for the company with the biggest name recognition and the most tech support. This unfortunate state of management is certainly a factor in all of this over-spending. If you really look at requirements it is very likely that you won't always need a IBM/BEA and their armies of tech support people. Sometimes you do, but make an informed decision, of course. Being a cynic (not without lots of examples and experience, mind you) I wonder if IT departments CAN make effective decisions.
  42. I would just like to point out that we are currently providing free download of an application server for developers - Rexip AppServer . It does almost everything except EJB. We believe this is a very compelling package for the 75% of the projects that Gartner refers to. With regard to Douglas Miller message, Rexip AppServer provides

    * SERVLETS ENGINE
    * JSP's SUPPORT (1.1/1.2)
    * SERVLET SUPPORT (2.2/2.3)
    * LOGGING
    * DB CONNECTION POOLING
    * SECURITY SUPPORT
    * CLUSTERING - using state-of-the-art in-memory replication with dynamic discovery
    * SESSION FAIL-OVER
    * THE BASE FOR ALL WEB APPLICATIONS

     In addition, the following features are also provided:

    * JMX-Based Remote Administration Tool
    * Watchdog for alerting administrators of abnormal server behavior\
    * Auto compiling for compiling and deploying source code automatically ot reduce development round-trip time
    * JavaMail 1.1
    * JNDI 1.2
    * JAF 1.0

    - Niles
    www.rexip.com
  43. Please provide a list of Production systems and articles were this app server is being used!!!!
  44. Well.

    "overspent $1 billion" not for everyone.

    at least that make app servers maker (IBM , BEA ....) a better living for the next years ...

     
  45. Well numbers in such reports can be deceiving in my view. When they say 60% of development will be JSP and Servlets, do they mean by number of projects, programs, lines of code or programmers? If it is number of programs then such a figure is not surprising if we consider the number of small sites out there or students experimenting in a lab somewhere!

    As was mentioned before in one of the comments the decision is yours. If your application requires the complexity of EJBs and all their "Goodness" (as does most of the work I do) then an app server is the choice.

    Their point is good one non-the-less only spend the money on an app server if there is compelling technical reason to do so!
  46. And even if they needed EJB or JMS functionality, they could use JBoss, Enhydra App Server, Rapid MQ and the list goes on.

    But is it not the developers who confuse their clueless managers to make these decisions?
  47. your damm right !!!

    my client don't feel safe if they don't spend money on "techologie" ....

    i use to ear " below 20k , this is not good techologie" .

  48. Using a big $$$ app server just for JSP/servlets is perfectly appropriate. We use HP Bluestone only for servlets and we couldn't live without it. Big app servers give you multi-process, multi-server scalability, session failover, hot versioning, etc.

    These features are worth every penny even if you don't get around to EJB.
  49. Posted by Jordan Zimmerman 2001-08-22 17:22:34.0.

    > Using a big $$$ app server just for >JSP/servlets is perfectly appropriate. We use HP
    > Bluestone only for servlets and we >couldn't live without it. Big app servers give you
    > multi-process, multi-server >scalability, session failover, hot versioning, etc.

    > These features are worth every penny >even if you don't get around to EJB.

    Big $$$ app servers may give you such functionality but you can also find some (if not all) of those features on cheaper app servers like Orion, Sybase et. al. And the thing is, these features will become more and more prevalent on some of the cheaper servers as the specs settle down and developers can concentrate on tweeking their products. Big $$$ app servers give you Product Name recognition for management and investors, Tons of tech support for developers and support staff - but they don't always have more (needed) features than some of the high quality but cheaper app servers.
    Cheers
    Ray
  50. We bought Bluestone in early 2000. There were no Open Source or low cost app servers that had the scalability feature set we needed at that time. The landscape is different now.
  51. BTW, which low cost or open source servers support clustering features? I just did a quick survey on the net and there aren't many that I can find.
  52. Rexip AppServer is your choice. It includes Clustering w/ load-balancing & failover (state-of-art im-memory replication), and many more... Detailed pricing will be released in early September, but I can guarantee you that it will be very competitive.

    If you are interested, would you please REGISTER as a member of our site. I assure you that I will keep you posted.

    - Niles
    www.rexip.com

  53. Niles,
    What is the price range? How does Rexip AppServer compared with JRun Pro, ServletExec and Websphere standard, all priced under US$1000.

  54. $1 billion "wasted" on app servers is chump change compared to tens to hundreds of billions wasted on CTOs' and VPEs' salaries, the same people who made the bad decision on buying a $15,000 ejb container to serve up static HTML pages. And even $15,000 can't buy you the expensive executive golf retreat packages awarded to these same folks for making the bad decisions in the first place!

    Face it, the $1 billion dollar is either going into IT deparments or the bonus-pools of non-deserving executives. I'll take a share of this ANYDAY!

    Gene

     
  55. I've seen the app server matrix on flashline <http://www.flashline.com/Components/appservermatrix.jsp>
    and searched far and wide for a real comparison; but, I've come up empty handed.

    Does anyone know where to find a rigorous cost/benefit analysis of the various app servers? In particular, I'm thinking about the explicit difference between using WebLogic versus an open-source app server. But, none of the analyses I've seen treat the extra developer's salaries one has to include for open-source support, or the cost to the architecture when the app server vendor goes out of business or radically changes the API (as was the case with netscape), etc. I'd love to see a true analysis of, say, WebLogic vs. Enhydra. Including *both* the feature/function differences and, more critically, the cost of implementation and maintenance.