Hi all I’m playing around orientdb to evaluate his inclusion into a new project.
Here is my problem.
Looking at the use cases I’m going to have a lot of super nodes (nodes which will have at least 5-10k outgoing relations) and I think that those nodes could be an irritating bottleneck on highly concurrent access.
The entire database must serve 20 departments, every department owns a partition of the data and those “blocks” are not accessible from the other departments.
Every department’s partition share about the 60% of the data structure schema and the other 40% of the schema is department independent…
At system level I have a couple of agents which have complete read access to the graph for data analysis and profiling and every department can have is own profiling agent which will profile only his partition data.
Now. My question is
Is possible to create “indipendent” sub graphs into an orient graph database?
Thanks to everybody for the time and the help.
Marco
You can model this use case inside your domain as graph:
root -> * departments -> other nodes
In this way department cross the graph starting for own department node.
To use something already done look at this post by Marko Rodriguez (the main author of Blueprints and Gremlin language): http://thinkaurelius.com/2012/04/06/multitenant-graph-applications/
And this recent project to run a partition graph on top of OrientDB‘s Blueprints implementation: https://github.com/tinkerpop/blueprints/wiki/Partition-Implementation