Discussions

News: Continuent renames C-JDBC to Sequoia with new release

  1. Sequoia v2.3 is the new release of the Continuent database clustering middleware. Sequoia is the same code base as the ObjectWeb C-JDBC project but licensed under an Apache License.

    Sequoia brings new features such as an Eclipse plugin and an ODBC driver for non-Java applications. It also breaks out the group communications layer in a project called Hedera.
  2. why rebranding ?[ Go to top ]

    Why is that a rebranding of C-JDBC (LGPL, on ObjectWeb)?
    Does that mean it is a fork? Or does it simply mean that to raise the visibilty some new names where choosen?

    From the web site

    Sequoia is our main project. It is the continuation of the C-JDBC project.

    It seems that C-JDBC is dead or abandonned? Is that the case? If so is Sequoia backed by ObjectWeb?

    I am quite surprised, though I knew there was a small business going on to offer services around C-JDBC.
  3. why rebranding ?[ Go to top ]

    Hi Alex,
    Why is that a rebranding of C-JDBC (LGPL, on ObjectWeb)?Does that mean it is a fork? Or does it simply mean that to raise the visibilty some new names where choosen?

    As Paul mentioned, we had to change the name because of Sun's trademark on JDBC. As the name is also present in the package name, it means that the code bases are different between Sequoia and C-JDBC.
    As for now the LGPL code base (C-JDBC) remains on ObjectWeb and the Apache licensed version (Sequoia) is hosted on Continuent.org.

    Emmanuel
  4. Congratuations, Emmanuel! I notice that Continent is open about the fact that it is using the C-JDBC 2.0 branch as a base for Sequoia, yet the license for Sequoia is the Apache License 2.0. Will ObjectWeb C-JDBC also be relicensed or dual licensed or is this change specific only to the Sequoia project? Also, what are the differences between ObjectWeb C-JDBC and Sequoia?
  5. Based on my understanding, the name change was due to infringement of the JDBC trademark (Sun's). It sounds like they were under pressure to abandon the use of JDBC as the project title. But I could be totally wrong about this.

    Anyway, I've checked on this project over the last few months, and am quite impressed at how far it has come. According to their literature, they have 8 full time developers working on the project now. Definitely seems pretty solid.
  6. Congratuations, Emmanuel! I notice that Continent is open about the fact that it is using the C-JDBC 2.0 branch as a base for Sequoia, yet the license for Sequoia is the Apache License 2.0. Will ObjectWeb C-JDBC also be relicensed or dual licensed or is this change specific only to the Sequoia project? Also, what are the differences between ObjectWeb C-JDBC and Sequoia?

    But since it relies on JGroups; its always gonna depend on LGPL code if thats what you were thinking Bruce :)

    So we can't use any of this stuff on Apache projects :(

    James
    LogicBlaze
  7. But since it relies on JGroups; its always gonna depend on LGPL code if thats what you were thinking Bruce :)So we can't use any of this stuff on Apache projects :(

    just write an Active* replacement for JGroups! Sequoia already talks to an abstraction layer, so its not directly tied to it.. :)

    -pete
  8. But since it relies on JGroups; its always gonna depend on LGPL code if thats what you were thinking Bruce :)So we can't use any of this stuff on Apache projects :(
    just write an Active* replacement for JGroups! Sequoia already talks to an abstraction layer, so its not directly tied to it.. :)-pete

    Actually, ActiveCluster (http://activecluster.codehaus.org/) is being designed to meet this need.

    Yet another wonderful project from ActiveStrachan ;-).
  9. Congratuations, Emmanuel! I notice that Continent is open about the fact that it is using the C-JDBC 2.0 branch as a base for Sequoia, yet the license for Sequoia is the Apache License 2.0. Will ObjectWeb C-JDBC also be relicensed or dual licensed or is this change specific only to the Sequoia project? Also, what are the differences between ObjectWeb C-JDBC and Sequoia?
    But since it relies on JGroups; its always gonna depend on LGPL code if thats what you were thinking Bruce :)So we can't use any of this stuff on Apache projects :(JamesLogicBlaze

    This is exactly what I discovered upon digging into Sequoia further yesterday. :-(

    You read my mind, James! ;-)
  10. But since it relies on JGroups; its always gonna depend on LGPL code if thats what you were thinking Bruce :)So we can't use any of this stuff on Apache projects

    As it was mentioned in the other posts, we have Hedera (http://hedera.continuent.org) that is a wrapper that abstracts the group communication. We already have Appia (http://appia.di.fc.ul.pt/) as a replacement for JGroups but it is LGPL as well. Any other group communication should be easy to wrap since the API is very simple (basically send/receive).
    In our commercial products (like p/cluster), we have our own proprietary group communication optimized for Sequoia.

    The other LGPL dependency is Octopus (ETL tool to backup/restore databases trhough JDBC) but you can replace it with your own backuper.

    Emmanuel
  11. Hi Bruce,
    Will ObjectWeb C-JDBC also be relicensed or dual licensed or is this change specific only to the Sequoia project? Also, what are the differences between ObjectWeb C-JDBC and Sequoia?

    ObjectWeb C-JDBC currently remains under LGPL and there are no plan to change the license yet.
    The main difference between C-JDBC and Sequoia is that the new developments and fixes are performed on Sequoia which is the code base we are working on. Fixes and improvements are backported to C-JDBC as resources and time permits. For the group communication C-JDBC uses the Tribe wrapper whereas Sequoia uses Hedera. They are functionally equivalent but we have groups working on replacement for JGroups to work with Hedera.
    The new developments (ODBC driver, Eclipse console, ...) are only tested with Sequoia. They should work with C-JDBC but we did not test so far.
    In terms of infrastructure we use JIRA on Continuent which greatly helps in the bug tracking compared to the Forge tracker.

    Hope this clarifies things a bit,
    Emmanuel
  12. Hi Bruce,
    Will ObjectWeb C-JDBC also be relicensed or dual licensed or is this change specific only to the Sequoia project? Also, what are the differences between ObjectWeb C-JDBC and Sequoia?
    ObjectWeb C-JDBC currently remains under LGPL and there are no plan to change the license yet. The main difference between C-JDBC and Sequoia is that the new developments and fixes are performed on Sequoia which is the code base we are working on. Fixes and improvements are backported to C-JDBC as resources and time permits. For the group communication C-JDBC uses the Tribe wrapper whereas Sequoia uses Hedera. They are functionally equivalent but we have groups working on replacement for JGroups to work with Hedera.The new developments (ODBC driver, Eclipse console, ...) are only tested with Sequoia. They should work with C-JDBC but we did not test so far. In terms of infrastructure we use JIRA on Continuent which greatly helps in the bug tracking compared to the Forge tracker. Hope this clarifies things a bit,Emmanuel

    So how was the project relicensed from LGPL to AL when the LGPL prohibits exactly this? I notice that there are copyrights in some of the code from INRIA. Did INRIA sign over its copyrights to AmicoSoft/Emic/Continuent?
  13. So how was the project relicensed from LGPL to AL when the LGPL prohibits exactly this? I notice that there are copyrights in some of the code from INRIA. Did INRIA sign over its copyrights to AmicoSoft/Emic/Continuent?

    INRIA has now the copyright on the entire code base. INRIA has granted an Apache License of the code to AmicoSoft/Emic/Continuent, so the code is currently dual licensed (LGPL/AL). INRIA remains the copyright holder of both code bases at the moment (except for new files which are under the copyright of their respective authors).

    Does that make sense?
    Emmanuel
  14. So how was the project relicensed from LGPL to AL when the LGPL prohibits exactly this? I notice that there are copyrights in some of the code from INRIA. Did INRIA sign over its copyrights to AmicoSoft/Emic/Continuent?
    INRIA has now the copyright on the entire code base. INRIA has granted an Apache License of the code to AmicoSoft/Emic/Continuent, so the code is currently dual licensed (LGPL/AL). INRIA remains the copyright holder of both code bases at the moment (except for new files which are under the copyright of their respective authors).Does that make sense?Emmanuel

    Yes, that makes perfect sense. Thanks for the clarification, Emmanuel.
  15. I got time to make a very simple test based on Sequoia 2.2,which has the followin platform:
    1 jdk1.4
    2 use APACHE IBATIS dao framework as data access layer
    3 use junit to create a simple test class to call a dao class to retrieve some data from MYSQL database
    4 use junitPerf to do the load test
    5 all tests are done in my laptop only,P3 800,RAM 256

    the results seem very interesting,
     the overall performence with sequoia 2.2 is kind of poorer than with direct jdbc connection!I guess the reason is calling sequioa is an extra resource usage of CPU and RAM.
    so,question is:
     how can i do tests using seqoia to achieve predictable result(as sequoia said, better performence) on one machine only?
    Thanks
    weidong
  16. all tests are done in my laptop only,P3 800,RAM 256

    Obviously if wou want to achieve scalability, there is no magic you need more hardware. Sequoia performs load balancing on queries (also called inter-query parallelism) but it does not speedup the execution of a single query by splitting it in smaller queries (this is intra-query parallelism).
    Also note that replication in general does not speedup write performance but mostly reads. That kind of solution is really meaningful if you need high availability or if you want to absorb peak loads that cannot be handled by a single machine.

    Emmanuel