Fluidinfo has been designed to scale dynamically. All aspects - storage, organization, and query processing - have been designed and built in a modular fashion, as services in a message-passing architecture. Storage is tag-oriented. This results in something that resembles a column store in which tags are fully independent of one another. Fluidinfo has been designed to make simple queries fast, and to eliminate query complexity by using an extremely limited query language whose execution can be trivially parallelized.
Although storage and computation within Fluidinfo are highly distributed, its administration and organization is centralized. There is a strong reason for this: Fluidinfo is principally concerned with making it possible to combine, augment, and search across information coming from multiple sources. That’s the whole point.
So for the time being, the coordination of Fluidinfo is centralized. We have of course given thought to federalization and have some ideas as to how that might be done.
Fluidinfo relies in part upon the following open-source components:
We plan to open source Fluidinfo, though the timetable for doing so has not been set.
We contributed Twisted support to the Thrift project and we are part of the Thrift team, an Apache Incubator project. We have also open-sourced txAMQP. txAMQP has been adopted by a number of other projects, some of which you can easily find via your preferred search engine.
Hopefully you now have a good understanding of what makes Fluidinfo special. Here a few suggestions to continue exploring Fluidinfo.