The two main use cases they have been designed for (although I'm sure there are others) are:
1. Infinitely scalable multi-tier data via a database.
Using two (or more) Omniscopes, with the first showing aggregated data, and the second showing an on-demand query of the selected data from the first Omniscope, but in a more granular form. This allows you to explode any in-memory limits on a given PC. For example, the first file might have a database of 1 million stocks, and the second would have your selected subset of those stocks' historical price data. Each stock might have 10,000 prices. 10 billion records is beyond the capacity of Omniscope, but selecting up to 100 stocks in one Omniscope, and viewing their prices in another, never exceeds 1 million records per Omniscope. http://forums.visokio.com/discussion/comment/2188/#Comment_2188
2. Automating a DataManager workflow with different parameters.
In Omniscope, you link various parts of your workflow (such as Record Filter or SQL filters) with parameters. You then use the Server to automate executing the workflow after changing the parameter values. See the server action "Set DataManager parameter value". You can do this either by configuring a schedule using the Scheduler, or by programmatically invoking the action using XML actions. For example, you wish to automate the publishing of different combinations of data sources and record subsets for different clients. (Note that there is overlap between this and the "Batch publisher" output block. Some behaviour may be easier using one or the other.)
Notes
Note that only some properties in DataManager are "parameterised", i.e. available for linking to parameters. If you find you need others that are not yet customisable, please post in the forums.
To see parameterised properties, you must have added at least one parameter in the data bar. After this, you will see a little "interlinked circles" icon next to the option, e.g. in the Record Filter block.