Discussions

News: Case Study: National Healthcare System of Brazil and Java

  1. Sun has posted a case study on how the National Healthcare System of Brazil is using Java and XML. The case study talks about a project to implement a national repository of health records.

    The interesting part of the architecture is that it is very old school, in that they just use JDBC. No Entity Beans, or JDO/Hibernate, or fancy DAO builder. The same goes for JAXP vs. JAXB/XML Beans/<insert cool XML system>, and for Servlets vs. Struts/WebWork/Faces/Tapestry/<insert cool web application framework>. This is probably just due to the age of the application though ("JAXB was not available at the time of the pilot").

    View the case study on: How Java Technology and XML Are Improving Healthcare in Brazil

    Threaded Messages (20)

  2. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
    The interesting part of the architecture is that it is very old school, in that they just use JDBC. No Entity Beans, or JDO/Hibernate, or fancy DAO builder.
    ============================================================

    Sun is writing a paper to market their technology not yours or your buddies. Why you are pointing to old technology? Smart people can build excellent application using simple JDBC
  3. Smart people can build excellent application using simple JDBC


    you don't need to be smart to build an excellent application with only JDBC, but to select it.
  4. I remember James Gosling talking about this in early 2003. See how he saw the system comparing to Sun's case study

    http://weblogs.java.net/jag/page1.html#8
  5. In a good way...[ Go to top ]

    I meant this in a GOOD way. They developed an app that works, and didn't just talk about buzzwords and bleading edge tech.

    Dion
  6. KISS[ Go to top ]

    Perhaps Vic Cekvenich is involved?

    Regards
    Rolf Tollerud
  7. KISS[ Go to top ]

    Perhaps Vic Cekvenich is involved?


    are they using struts and ibatis? :)
  8. Makes me proud[ Go to top ]

    This makes me even more proud of being a Brazilian software developer. Other countries should take a look at our electronic voting system too, the most advanced in the world:

    http://www.enterbrasil.com/election/election.html

    The only bad thing about it: it's not based on Java technology... :(

    Regards,
    Henrique Steckelberg
  9. Makes me proud[ Go to top ]

    Other link about the project:

    http://www.saude.gov.br/cartao (in Portuguese only)

    Regards,
    Henrique Steckelberg
  10. Makes me proud[ Go to top ]

    I know we've looked at your bus system (Curitiba).
  11. I'm very proud of this (healthy system) news too. As a brazillian developer, I'm sure that we one of the most advanced coutries in Java. I work in the Electoral Court and think that it could use more open technologies, as linux and Java, and the lack of a more open-source sight of the Electoral big heads..
    But nobody can't say brazil hasn't one of the most success eletronic vote system ever done. In this year (as in 2002) we'll have all the cities covered by the Urna Eletrônica (as we call the vote machine), and that's not small think, as there's a lot of remote places hard to get to.

    Emerson Cargnin
  12. This is good stuff. KISS is a pattern in itself, and probably the most neglected in business architecture.

    The Sun article also does justice to this principle. It is clearly laid out and easy to understand, and manages to be fairly comprehensive without sacrificing too much technical detail. All in 10 pages! An object lesson in how to construct a high-level architecture document.
  13. KISS[ Go to top ]

    No, KISS is not a pattern, it's a goddamned slogan.
  14. KISS[ Go to top ]

    No, KISS is not a pattern, it's a goddamned slogan


    Actually it's the one of the best pieces of advice you'll ever get as a software engineer.
  15. KISS[ Go to top ]

    Actually it's the one of the best pieces of advice you'll ever get

    > as a software engineer.

    You can't measure it. Most people will not agree on it when they see it.
    Most people will not agree on a way to get it. There's more than one
    way to get there. It's very personal. It's very context sensitive.

    It's very sloganish because it is ineffable. It's more of guise
    for everyone to do what they want under the protection of a
    lofty principle.
  16. KISS[ Go to top ]

    When I say that "KISS is a pattern in itself", I am thinking in terms of Christopher Alexander's original definition, of which context sensitivity is indeed part. IIRC he defined a pattern as a three part construct:

    (1) the context (i.e. under what conditions does this pattern hold).
    (2) a system of forces that exert pressure away from a problem and towards a solution or goal.
    (3) the solution itself; a configuration that balances the system of forces or solves the problems presented.

    I would therefore say that KISS meets the criteria of pattern identity, since:

    (1) it holds under conditions where simplicity does not degrade quality, and where additional complexity does not provide appreciable and forseeable benefit.
    (2) there is a system of forces (usability, maintainability, testability, understandability, etc.) which exert pressure away from complex solutions and towards simple ones.
    (3) reduced complexity (fewer artifacts, nodes, dependencies, SLOC) balances those forces, and solves those problems concerning usability, maintainability etc.
  17. The amount of data.[ Go to top ]

    "> 50 gigabytes of data/million users/year"

    Our clinic is producing one terabyte of radiology data every 6-7 months for 10000 -15000 patients. Actually we are working on the same project in Germany and it is very interesting for me how they solved problem of transfer such a big data to the central servers.
  18. The amount of data.[ Go to top ]

    "> 50 gigabytes of data/million users/year"

    >
    > Our clinic is producing one terabyte of radiology data every 6-7 months for 10000 -15000 patients. Actually we are working on the same project in Germany and it is very interesting for me how they solved problem of transfer such a big data to the central servers.

    It is very interesting project, I hope you do not use XML, do you ?
  19. The amount of data.[ Go to top ]

    The radiology data stays in original DICOM format and we use streaming to deliver DICOM data to the end points. The rest data is in XML.
  20. The amount of data.[ Go to top ]

    The radiology data stays in original DICOM format and we use streaming to deliver DICOM data to the end points. The rest data is in XML.


    I am not expert and there is no way to implement something like this in our country (internet connection is too slow or it can be more expensive than data itself). We have two ways, reduce and compress data, drop XML or to use bag to transport this kind of data if compression does not help.
  21. The amount of data.[ Go to top ]

    I am not expert and there is no way to implement something like this in our country (internet connection is too slow or it can be more expensive than data itself). We have two ways, reduce and compress data, drop XML or to use bag to transport this kind of data if compression does not help.


    We didn't solve data transfer problem finally, but for end users, e.g. for doctors we deliver radiology data through streaming and it works fine even with slow modem connection.
    Actually the model of our application is different than one in Brazil, but problems with data transfer are same.