Application Cloud clustering

As a system admin I would like to be able to deploy geonetwork to cloud based services in a dynamic way having all the required data accessible to the new instance as it is spun up and be certain that no data will be lost when the system is turned off.

The metadata resources, database, index, configuration, etc... must all be stored in such a way that it is accessible to all instances (when applicable) editable when required and not tied to a particular instance (in case of crashed shut down of dynamically started services)

Other considerations are

• the distinction between read-only servers (maybe RegisteredUsers and public can be redirected to use those servers)

• Read/write servers for editors and up

• How are do the admin tools like settings, harvesters, etc... spread across the instances in the cloud

• Would harvester execution be sharded across the cloud?

• If harvester execution is sharded, what happens when an instance is shutdown

• What tool would be used to coordinate between instances

• etc...

• Use Path backed by DB for scaling so all data goes to DB and can be shared, duplicated, etc... Maybe sharded?

• Thesaurus is bound to disk because of the RDF library we are using we need to move to a new thesaurus technology

• Keeping index in sync across all instances can be difficult.

• Maybe put full MD in solr and rendering only from solr so read-only doesn't use DB. Only writes/updates uses DB (except for download resources)