ZK Clustering

answered 2010-01-06 16:58:42 +0800

Phileas
171 ● 3

By the way, we have in these pages ZK node clustering with only one apache, what is possible with multiple apache ?

answered 2010-01-07 10:41:30 +0800

Phileas
171 ● 3

So it's really ZK clustering, not web application clustering as if apache goes down, all is down...
Except if my idea is wrong but if we have an integration between ZK and Terracotta middleware, that is a ZK TIM (Terracotta integration module), then that should gives us ZK web application clustering.

Terracotta people are open to collaboration with other project owners to work hand in hand so i'm pretty sure they would be glad of ZK responsible interest by proposing the ZK added value into their product.
On a position point of view, it would be a winner/winner relation in terms of added value for both community and each product popularity.

answered 2010-01-08 13:30:46 +0800

Phileas
171 ● 3

Hello Terrytornado, tmillsclare,

@Terry : i've just been told that except for the very first chapters on general informations, the book is already outdated and the new updated one will be out in the very soon future.

@Tmill : Do you mean you work in the ZK Team ? or will you join it by your efforts ?

answered 2010-01-28 17:40:46 +0800

Phileas
171 ● 3

Thanks PeterKuo,

As mentioned on the Tomcat forum thread : If for any reasons, that relationship or whatever does not work as desired, there are also complete free alternatives :

- Hazelcast http://www.hazelcast.com/product.jsp
- Shoal https://shoal.dev.java.net/

answered 2010-01-29 16:14:37 +0800

Phileas
171 ● 3

updated 2010-01-29 17:11:05 +0800

Thanks for the kind words terrytornado,

Now, if ZK team continues its interest into Terracotta (or by side effect some alternatives), here are some potential others resources bringing their added value :
Jonas Bonér worked at Terracotta and is the founder of Akka (http://akkasource.org/) for which Scala language is used (the well known concise and elegant language over the JVM which smoothly integrates features of object-oriented and functional languages). Akka, on its side relates to writing correct concurrent, fault-tolerant and scalable applications essentially by using the Actor Model together with Software Transactional Memory .

Enough marketing :) , what is interesting is the work of Jonas about "Clustering Scala Actors with Terracotta" (presentation here http://jonasboner.com/2008/01/25/clustering-scala-actors-with-terracotta/).
My second level of my interest into Akka (http://akkasource.org/) is the "simplicity" approach of the whole subject Akka is made for, along with excellent modules we can plug like Shoal i've just listed above and NOSQL approach when the subject comes to scalable performance and flexible data model (Cassandra key-value db, MongoDB document based or soon Neo4j graph db ... or even the Terracotta Store which is also document based).

Great goals, simplicity and performance are things i like to think about when the question comes to mix ZK with "companion products" :), that's why i would suggest to consider it as an alternative way to answer objectives Terracotta and these others products (which are in the Akka "culture") address. It seems Akka team worked "less" on the web presentation integration by supporting "only" the Lift framework, thinking here ZK would perfectly suit the place it could have with Akka, something like one turbo bullet, several prizes :)