Max Rydahl Andersen wrote:
> Recently I've been working on Hibernate Search clustering.
> Hibernate Search will have the ability to work in a master / slave mode.
>
> Every node has a local copy of the index and can read it (ie query it).
> When an update is required (actually when a transaction scoped list of
> updates are required) by a node, a message is sent to a master node
> responsible to update the master index.
> On a regular basis, each node acquires a copy of the master index in
> local to stay up to day asynchronously.
>
> My first implementation involves JMS, but one could imagine using:
> - JBoss Messaging standalone in headless mode (when it's done)
> - a servlet controller receiving http requests
> - whatever you want
>
> Thoughts?
Is this clustering at the level of Hibernate Search or is it something
at the lucence index level ?
Hibernate Search
Would it not make sense to be a "index tech" level concern or ?
No because the index has no notion of transaction and batch of operations.
However I plan sometime to create a JBossCache Directory (index
container). The scalability is limited though since Lucene needs an
index based locking system.
--
--
Max Rydahl Andersen
callto://max.rydahl.andersen
Hibernate
max(a)hibernate.org
http://hibernate.org
JBoss a division of Red Hat
max.andersen(a)jboss.com