Skip to content

201311clustering

Jump to bottom
Jeroen Ticheler edited this page Nov 22, 2013 · 21 revisions

201311clustering_old

#Scalability and Clustering

When SOLR is used for indexing, scaling across multiple instances of GeoNetwork requires SOLR and DBMS synchronization. The data directory can be shared across the different nodes.

https://raw.github.com/geonetwork/wiki_images/master/development/cluster.jpg

Java Message Service (JMS) is possibly still required to synchronize information loaded in memory. It should be investigated if other strategies exist to keep memory in synch across nodes when information is updated in one of the nodes. For example when a thesaurus is edited.

In the JMS section below we describe what information is most likely affected.

##JMS

In existing implementations Java Message Service (JMS) is used to synchronize the following information across a cluster of GeoNetwork nodes:

  • Indexing
  • Settings & System configuration
  • Thesaurus
  • Harvesters
  • Metadata versioning

Detailed proposal implemented using JMS and Lucene indexes: http://trac.osgeo.org/geonetwork/wiki/LoadBalanceable

Interest Bernhard, Ine, Paul, Maria, Lars

Clone this wiki locally