How does it work?

The OMII Resource-Aware Visualization Environment (OMII-RAVE) client contacts a predetermined Grimoires server (such as that hosted by the NGS), and searches for OMII-RAVE data services. Each such service is then interrogated for currently hosted datasets, which the user may instruct the client to subscribe to.

If the user to upload a new dataset (to be advertised to and discovered by other users), the client will scan available data services and upload the dataset to the most appropriate data service (through various heuristics, such as machine load, available memory, network bandwidth, etc.).

If a user instructs the client to collaboratively view a dataset, the client verifies if the local machine has sufficient resources to host the dataset locally; if not, an OMII-RAVE render service is selected (via heuristics). This render service will then download a copy of the dataset on behalf of the client, render views of it and return the rendered images as a movie stream to the client for local display (and interaction). If the dataset is hosted locally or remotely, the user does not need to know, nor is there any difference in the locally displayed GUI.

Note that a single data service or render service can host multiple datasets, each dataset in turn hosting multiple simultaneous users. This supports resource sharing, and enables collaboration through multiple people viewing the same dataset simultaneously. Each simultaneous user is depicted in the dataset with a simplified avatar, so their location (and view direction) is known to all, enabling such concepts as guided tours or to publicise an interesting viewpoint that they wish to share.

Each dataset is hosted in a newly spawned JVM for additional security; if any aspect of the service suffers a critical failure, it will not affect any other hosted datasets. Datasets are hosted for a given lifetime, after which the dataset is discarded and the resources returned for use by other datasets on the server machine.