| 32 | |
| 33 | == Example == |
| 34 | |
| 35 | The diagram below illustrates a possible cluster configuration. The cluster consists of 3 redundant GeoNetwork instances. The instances share the database, the shared data directory, and the JMS server ([http://activemq.apache.org/ ActiveMQ]). A loadbalancer hides the actual, different URLs to these GeoNetwork instances so this clustered catalogue is accessed by users through a single URL. The users do not know which GeoNetwork node serves them, nor that this catalog is horizontally scaled at all. |
| 36 | |
| 37 | [[Image(cluster.png)]] |
| 38 | |
| 39 | You notice that each of the GeoNetwork nodes maintains its own Lucene index and (if configured) SVN repo. This does not impair functionality as the indexes and the SVN repos are kept in sync. However it might be better to use a shared Lucene index, at least. Future work may explore how to do that, possibly using [http://lucene.apache.org/solr/ SOLR]. |
| 40 | |
| 41 | Note that the database and the JMS server could be horizontally scaled as well, but that's outside the scope of this proposal. |
| 42 | |