Taking a Look at Version 2.1 of Objectivity’s InfiniteGraph
Paul Williams writes:
InfiniteGraph is a distributed graph database application developed by the California-based company, Objectivity. Companies focused on the relationships within their data make up the primary market for InfiniteGraph. The database is known for its ability to find connections inside large datasets, as well as its robust performance and easy scalability.
InfiniteGraph uses a unique load-based pricing model that allows interested parties to try the software, including full database development, essentially free of charge. Two pricing options exist for companies deciding to fully deploy InfiniteGraph. The first option is “pay as you go,” which sports a run-time usage-based pricing model. Companies with large or classified applications can take advantage of a site-wide license option.
Those interested in using the free demo version of InfiniteGraph need to either have a Java compiler combined with some skill using a command line, or an installed IDE such as Eclipse. Database model development in InfiniteGraph requires at least some basic familiarity with the Java language, considering models and relationships (vertices and edges in InfiniteGraph nomenclature) are defined as Java classes. Once a database is compiled, the InfiniteGraph Visualizer app (included with package) allows for graph navigation and data browsing.
Objectivity might benefit from providing a pre-compiled database with the InfiniteGraph download package, so interested parties can investigate the Visualizer app and the software’s data mining capabilities without the need of a Java compiler and/or having to engage their development staff.
Just going off of the review for the moment, I think having a common navigation of data and metadata is a good thing. That is offset by the rather unnatural load a graph database and search for a node before there is a display.
I think I understand the reasoning for the search for a node first before displaying but it is counter-intuitive, where counter-intuitive = burden on user. Better to provide for (and use) a default node that is displayed upon load. That way “load” produces some action the user can see. And offer the user the ability to pick another “default” node that loads automatically when the graph is loaded.
It sounds like the documentation could be better integrated into the application (and not left on the vendor’s website).
Comments or suggestions on strong/weak points to look for with InfiniteGraph?