Discussions

News: JXInsight 5.6 Released

  1. JXInsight 5.6 Released (14 messages)

    JXInsight 5.6 has been released with significant performance enhancements in its open and extensible (multi-meter) resource metering runtime. Using SPECjvm2008 as a benchmark for Java performance profilers JXInsight 5.6 is 6-50 times faster than comparable solutions. The free development edition of JXInsight 5.6 can be downloaded from: http://www.jinspired.com/products/jxinsight/downloads/index.html A more detailed summary of the enhancements is available here: http://www.jinspired.com/products/jxinsight/new-in-5.6.html Further information related to the new enhancements are available on the company blog: http://blog.jinspired.com/?cat=63

    Threaded Messages (14)

  2. Benchmarks[ Go to top ]

    "Using SPECjvm2008 as a benchmark for Java performance profilers JXInsight 5.6 is 6-50 times faster than comparable solutions." That's a big boast but considering the site doesn't specify who the competitors were, what app was being benchmarked, nor how they or JXInsight were configured for the benchmark t just makes it sound like something the marketing team came up with. I expect this for apps used by consumers but not for developer apps (i.e., don't treat us like we're stupid please).
  3. Re: Benchmarks[ Go to top ]

    Hi Marc, Maybe you forgot to take into account the following before firing off a rash and rather empty reply... 1. Many commercial software vendors prohibit publication of benchmarks related to their software offerings. 2. SPECjvm2008 is an official benchmark for Java runtimes and is FREELY available. 3. JXInsight 5.6 is also FREELY available for development usage. 4. I made the claim and I do not publish BS. 5. You can do your own test of the claims. 6. The configuration has been published if you just took a moment to some small amount of research before posting. http://blog.jinspired.com/?p=247 In fact the claims in the original posting were ** incorrect ** the true comparison factoring out the actual benchmark processing itself and just comparing the raw measurement and instrumentation overhead of the products is 17-100 times. Yes we had understated how fast JXInsight Probes really is in a production environment. You might also like to know that tests with our initial next release under development indicate we can go even much faster once we remove some native dependencies and we are going to scale even better. We are not alone in thinking our probes approach is currently the most ideal solution for resource metering of both coarse grain and fine grain event intervals. William
  4. Re: Benchmarks[ Go to top ]

    "Maybe you forgot to take into account the following before firing off a rash and rather empty reply..." Empty like your unproven claims? Everything I wrote is true based on the information you provided in you initial post. If asking you to back up your claims makes me rash...then I'm guilty as charged! "I made the claim and I do not publish BS." Yes, I guess I should believe everything complete strangers trying to sell me something tell me. "The configuration has been published if you just took a moment to some small amount of research before posting." You post the invocation parameters of JXInsight only. Also, my reply was based on looking at the links YOU provided. "You can do your own test of the claims." Why should I spend my own time testing YOUR claims? Either back up our claims or don't make them. Sorry but if you post stuff like this to a developers blog and not expect to get called on it, you're fooling yourself. Products and opinions get critiqued on this site every day. It's a good thing that inspires debate and helps cut through the BS.
  5. Re: Benchmarks[ Go to top ]

    Yes, I guess I should believe everything complete strangers trying to sell me something tell me.
    I am a complete stranger on TSS? LOL. William
  6. Re: Benchmarks[ Go to top ]

    hmm, what has happened to the good old "the customer is king" principle?
  7. Re: Benchmarks[ Go to top ]

    Don't fret about it Marc. Pretty soon you realise that no-one actually uses JXInsight apart from William ;)
  8. Re: Benchmarks[ Go to top ]

    Wow We're really hostile here :) Hope you have a thick skin Will. Anyway, now for my turn at being flamebait. I'm happy to say that Will added support for WebSphere eXtreme Scale aka ObjectGrid to V5.6 also. Billy (yes, I work at IBM, yes WebSphere eXtreme Scale is the product I work on).
  9. Re: Benchmarks[ Go to top ]

    Anyway, now for my turn at being flamebait.
    Hehe. Caught me out. Was hoping to get a bite from Will though.
  10. Re: Benchmarks[ Go to top ]

    Come on Mark. You have to be less nit picky and look at the big picture. The benchmarks don't really matter because by next month JXInsight will be so fast you'll need a nitrogen cooled JVM just to run it :-). You can always take a look at JEXIN as an alternative. It's not actually used for the same purpose as JXInsight but William hijacked my release post to get some free advertising by comparing apples and oranges so it seems only fair that I do the same.
  11. Re: Benchmarks[ Go to top ]

    Come on Mark. You have to be less nit picky and look at the big picture. The benchmarks don't really matter because by next month JXInsight will be so fast you'll need a nitrogen cooled JVM just to run it :-).

    You can always take a look at JEXIN as an alternative. It's not actually used for the same purpose as JXInsight but William hijacked my release post to get some free advertising by comparing apples and oranges so it seems only fair that I do the same.
    +1 Would you care to comment on any of the features coming in your next release?
  12. Re: Benchmarks[ Go to top ]

    Would you care to comment on any of the features coming in your next release?
    Sure. Here is what I'm considering for the next release based on feedback:
    • Support for injecting checked exceptions
    • Detecting injectable exceptions from the throws clause of traceable methods
    • Support for Spring defined aspects so you don't have to use AspectJ (AspectJ will still be supported)
    • An integration point for users to implement their own paradigm for deciding what methods are traceable
    • Remove the requirement that traceable exceptions be annotated with @Traceable
    Ideas for releases after the next one include:
    • Documentation on runtime weaving to add traceability to third party libraries
    • Ability to examine/alter method return values
    • Ability to define exceptions to inject by entering the injection class name directly in the UI.
    • Persistence for nodes, stack traces, and templates
    Please feel free to request features using jexin'sfeature request tracker. That's my primary consideration when deciding what to implement.
  13. Re: Benchmarks[ Go to top ]

    Come on Mark. You have to be less nit picky and look at the big picture. The benchmarks don't really matter because by next month JXInsight will be so fast you'll need a nitrogen cooled JVM just to run it :-).

    You can always take a look at JEXIN as an alternative. It's not actually used for the same purpose as JXInsight but William hijacked my release post to get some free advertising by comparing apples and oranges so it seems only fair that I do the same.
    +1 Actually +100 That last round of flames from Will removed JXInsight from my radar.. I'd hate to see the treatment I'd get if I called in for tech support..
  14. Congrats on the release, William. All I have to say is that JXInsight is a wonderful product. Our users are very happy with the integration it provides with GridGain. Best, Dmitriy Setrakyan GridGain - Grid Computing Made Simple
  15. Damn it! Too bad there is no way to edit a post on TSS. Anyway, here is the link to JXInsight/GridGain integration. Integration JXInsight/IBM eXtreme Scale looks as good by the way. Best, Dmitriy GridGain - Grid Computing Made Simple