POJO Mojo

Terracotta 2.1 has been released and is available for download. This release introduces a new product called Terracotta for Spring (Release Candidate 1), made specifically for clustering web applications deployed on the Spring Framework. Terracotta for Spring includes support for clustering JMX, Web Flow, and singleton Spring Beans. Additionally, support for SUSE Linux 10 and Solaris 10 has been added to Terracotta Sessions and Terracotta DSO, and the release contains a number of performance enhancements such as new pre-fetching heuristics to predict which objects a node will access.

Yes, its true. Our software is now downloadable without your needing to register. If you want to run a Spring app that shares state w/o Hibernate. If you want to cluster session underneath a Struts, Spring Web Flow, or other conversational web application but cannot stand Java serialization. If you want access to massive datasets from thousands of servers. If you want these capabilities without RMI, JDBC, JGroups, filesystem sharing or custom clustering, then Terracotta can help.

Terracotta's software simplifies JAVA scalability and availability by clustering your application's heap. Try it out, experience it for yourself. And you can contact us if it works for you and you start thinking about moving to production.

I am very excited about being able to allow people to prove to themselves that it works without any hinderances. I am looking forward to all the questions and feedback out there.

If you need what we have, no need to wait anymore. Give it a shot.

Hung out at the Philly JUG last night:
http://duke.csc.villanova.edu/phillyjug/index.jsp

Those guys asked some great questions and really engaged on the topic of transparent clustering and stateful versus stateless application design. Interestingly enough, most of the 100 people in the room seemed to agree that stateless app design presents good operational characteristics. I can load balance using non-sticky (read: cheaper load balancing setup) and I can kill any app server at any time, unilaterally. Last, I can scale just by scaling my data store.

But, Serialization impacts, and OR-mapping can slow you down and that gets frustrating when you know the data you are storing really is only being saved BECAUSE you are stateless (truly transient data, but you want it around in case of server failure--like Spring Web Flow content or Struts' data insession).

Wouldn't you rather be "stateful" in your design too?

I can't wait till my new article on the topic gets posted up at DevX.

Thanks Philly!

This tutorial walks a developer through clustering a simple Swing application comprised of a slider. Take this step-by-step tutorial to find out how to use configuration-driven clustering technology, avoid custom APIs as well as Java Serialization, and set up Eclipse for doing clustered programming going forward.

I did it myself, from scratch, and it took 11 minutes. Give it a try.