News: Tech Talk with Dion Almaer on Clustering TSS, J2EE Portability

  1. In this interview, Dion discusses TheServerSide cluster project and what was involved in porting the TSS codebase to different application servers; he looks at how TSS added a distributed cache to share data between the servers, and how it was strapped into the entity bean layer. He also talks about the Java AVK, features recently implemented on TSS, and future directions for the site.

    Watch Dion Almaer's Interview

    Also read Give your DB a Break: Using Caching for Speed and Availability for information on the clustering/caching that TSS uses.

    Threaded Messages (12)

  2. Damn photo[ Go to top ]

    I like how Nitin tried to find the worse possible photo of me... but then I checked and it was the best ;)
  3. no worries...[ Go to top ]

    the content more than made-up for that bulldog-mug ;o)

    Seriously, don't sweat the photo. It looks fine.

    Thanks for the insight about the architecture and implementation. Non-lofty architects who still have the sack to go into code are a rare breed (it takes one to know one).

    Your comments gave me several ideas! Thanks again!

    John C. Dale
  4. Foto[ Go to top ]


    you are married...so the foto should be less good :-)

    I like your approach with TheServerSide...maybe it is not the coolest thing, but it works! You are running this site on many different servers of different vendors and the performance with so many users reading the articles is very good.

    We should never try to make a good or the best architecture but the right architecture.

  5. Damn photo[ Go to top ]

    Well, considering how many female collegues I have had over the years, I doubt that there will be a lot of "hot chicks" checking out this site for sex bombs. ;)

    And considering the looks of myself and most of my past collegues, I doubt any of us are sex bombs, unless that is by definition "how much pr0n you have on your harddrive" ;)

    Ok, this has got to be the most pointless post here in ages (including Rolfs), so just click away on the "noisy"-button..
  6. Distributed Cache[ Go to top ]

    I'm sure you're already aware of this, Dion, but the latest release of Hibernate (2.1 rc1) has an integrated clustered cache. This probably gives you a more elegant solution than hacking a distributed cache into BMP's, with n+1 Finders and all.
  7. Distributed Cache[ Go to top ]

    Corby -

    I totally agree that using something like Hibernate, or JDO with caches built in would be great (they can even plug into Coherence... as well as other implementations).

    The reason that we didn't do this, was that I had a very short time to get the cluster going, and it took a day or two to implement the BMP -> cache integration code (since the BMPs were already written). TheServerSide is legacy EJB code ;)

    We don't really run into problems like n+1, since everything is in the cache.

    However, I again agree that it would be a lot nicer to use a nice transparent persistence model with built in distributed caching. We will move to this in a new version of TSS.


  8. Distributed Cache[ Go to top ]

    We don't really run into problems like n+1, since everything is in the cache.

    Maybe this is the reason why there is no message editing option on the server side, as message editing would break the read only cache, ;)

    On some forums users can edit their own messages. This can be very usefull for correcting spelling errors, although this option could be missused.

  9. Distributed Cache[ Go to top ]

    to have a cache for read-only/mostly data is not a question. Could anyone share thoughts/experience on heavily read/write transactional data cache (for example an ebay item in last minute :)? What is better to read/write to database or just write and allow a cache to synchronise the data on changes?
  10. Distributed Cache[ Go to top ]

    Mileta: Maybe this is the reason why there is no message editing option on the server side, as message editing would break the read only cache, ;)

    Not at all ... the Coherence cache type that Dion is using is designed for highly concurrent access and modification, and can pretty easily support a thousand updating transactions per second. Switching to the automatically partitioning cache type, it would scalably handle several thousand transactions per CPU per second. It's anything but a read-only cache ;-)


    Cameron Purdy
    Tangosol, Inc.
    Coherence: Clustered JCache for Grid Computing!
  11. Editing messages[ Go to top ]

    Yes... although users can't edit their own messages, I often go in and edit Cameron's messages :) [just kidding!]
  12. PODF or doc file[ Go to top ]

    Can we also have a pdf or doc version of the interview as due to poor bandwith , we cannot see the video? Hence we are not able to gain form the information shared by Don & others
  13. Out of curiosity, which server could you not get SiteMesh to work on?